This paper demonstrates fundamental limits of sensor networks for detection problems where the number of hypotheses is exponentially large. Such problems characterize many important applications including detection and classification of targets in a geographical area using a network of seismic sensors, and detecting complex substances with a chemical sensor array. We refer to such applications as large-scale detection problems. Using the insight that these problems share fundamental similarities with the problem of communicating over a noisy channel, we define the “sensing capacity” and lower bound it for a number of sensor network models. The sensing capacity expression differs significantly from the channel capacity due to the fact that for a fixed sensor configuration, codewords are dependent and nonidentically distributed. The sensing capacity provides a bound on the minimal number of sensors required to detect the state of an environment to within a desired accuracy. The results differ significantly from classical detection theory, and provide an intriguing connection between sensor networks and communications. In addition, we discuss the insight that sensing capacity provides for the problem of sensor selection.

Sparse measurement structures - in which each measurement only depends on a small number of the inputs-arise in models of many problems such as sensor networks, group testing and even lossless data compression. The most important question in these applications is `How many measurements are sufficient to reconstruct the input ?'. Regular structures, where each input is measured the same number of times, require fewer measurements than when arbitrary measurements are used. In this paper we conduct a general analysis of the performance of these regular measurement structures when used for lossy reconstruction of the input in the presence of measurement noise. The main contribution of our work is the generality of the result which is applicable to lossy detection with arbitrary, even non-linear measurements and with inputs and outputs in any discrete domain. The second contribution is the quantification of the effect of noise in the measurements on the reconstruction performance, which can be used to analyze the robustness of these sparse measurement structures. We show the generality and applicability of our results by analyzing the performance of pooling designs for rare allele detection.

Seminal work by Chevillat and Costello showed that for specific convolutional codes transmitted over a binary symmetric channel and decoded by sequential decoding, a measure of decoding effort decreases exponentially with the column distance function of the code. This has led to a large body of research in the design of codes with good distance profiles which are also used for transmission over Gaussian channels. In this paper we analyze the computational complexity of a stack decoder working on a specific tree code with real (as opposed to binary) symbols, transmitted over a memoryless Gaussian channel. In contrast to prior work that used random coding arguments, we use the intuition provided by the original proof to prove that decoding effort exhibits similar behavior even for a memoryless Gaussian channel. Our result is applicable to convolutional codes with antipodal signaling, sequence detection over Gaussian ISI channels and some sensor networks.

The exploration problem is a central issue in mobile robotics. A complete coverage is not practical if the environment is large with a few small hotspots, and the sampling cost is high. So, it is desirable to build robot teams that can coordinate to maximize sampling at these hotspots while minimizing resource costs, and consequently learn more accurately about properties of such environmental phenomena. An important issue in designing such teams is the exploration strategy. The contribution of this paper is in the evaluation of an adaptive exploration strategy called adaptive cluster sampling (ACS), which is demonstrated to reduce the resource costs (i.e., mission time and energy consumption) of a robot team, and yield more information about the environment by directing robot exploration towards hotspots. Due to the adaptive nature of the strategy, it is not obvious how the sampled data can be used to provide unbiased, low-variance estimates of the properties. This paper therefore discusses how estimators that are Rao-Blackwellized can be used to achieve low error. This paper also presents the first analysis of the characteristics of the environmental phenomena that favor the ACS strategy and estimators. Quantitative experimental results in a mineral prospecting task simulation show that our approach is more efficient in exploration by yielding more minerals and information with fewer resources and providing more precise mineral density estimates than previous methods.

Prior work motivated the use of sequential decoding for the problem of large-scale detection in sensor networks. In this paper we develop the metric for sequential decoding from first principles, different from the Fano metric which is conventionally used in sequential decoding. The difference in the metric arises due to the dependence between codewords, which is inherent in sensing problems. We analyze the behavior of this metric and show that it has the requisite properties for use in sequential decoding, i.e., the metric is, 1) expected to increase if decoding proceeds correctly, and 2) expected to decrease if more than a certain number of decoding errors are made. Through simulations, we show that the metric behaves according to theory and results in much higher accuracies than the Fano metric. We also show that due to an empirically-observed computational cutoff rate, we can perform accurate detection in large scale sensor networks, even when the optimal Viterbi decoding is not computationally feasible.

Computing the exact maximum likelihood or maximum a posteriori estimate of the environment is computationally expensive in many practical distributed sensing settings. We argue that this computational difficulty can be overcome by increasing the number of sensor measurements. Based on our work on the connection between error correcting codes and sensor networks, we propose a new algorithm which extends the idea of sequential decoding used to decode convolutional codes to estimation in a sensor network. In a simulated distributed sensing application, this algorithm provides accurate estimates at a modest computational cost given a sufficient number of sensor measurements. Above a certain number of sensor measurements this algorithm exhibits a sharp transition in the number of steps it requires in order to converge, leading to the potentially counter-intuitive observation that the computational burden of estimation can be reduced by taking additional sensor measurements.

Abstract In an ad hoc network, intermediate nodes on a communication path are expected to forward packets of other nodes so that the mobile nodes can communicate beyond their wireless transmission range. However, because wireless mobile nodes are usually constrained by limited power and computation resources, a selfish node may be unwilling to spend its resources in forwarding packets which are not of its direct interest, even though it expects other nodes to forward its packets to the destination. It has been shown that the presence of such selfish nodes degrades the overall performance of a non‐cooperative ad hoc network. To address this problem, we propose a secure and objective reputation‐based incentive (SORI) architecture to encourage packet forwarding and discipline selfish behavior. Different from existing schemes, under our architecture, the reputation of a node is quantified by objective measures; the propagation of reputation is efficiently secured by a one‐way‐hash‐chain‐based authentication scheme; and secure routing is in place. Armed with the reputation‐based mechanism, we design a punishment scheme to penalize selfish nodes. The experimental results show that the proposed scheme can successfully identify selfish nodes and punish them accordingly. Copyright © 2006 John Wiley & Sons, Ltd.

We bound the number of sensors required to achieve a desired level of sensing accuracy in a discrete sensor network application (e.g. distributed detection). We model the state of nature being sensed as a discrete vector, and the sensor network as an encoder. Our model assumes that each sensor observes only a subset of the state of nature, that sensor observations are localized and dependent, and that sensor network output across different states of nature is neither identical nor independently distributed. Using a random coding argument we prove a lower bound on the 'sensing capacity' of a sensor network, which characterizes the ability of a sensor network to distinguish among all states of nature. We compute this lower bound for sensors of varying range, noise models, and sensing functions. We compare this lower bound to the empirical performance of a belief propagation based sensor network decoder for a simple seismic sensor network scenario. The key contribution of this paper is to introduce the idea of a sharp cut-off function in the number of required sensors, to the sensor network community.

Occupancy grid mapping algorithms assume that grid block values are independently distributed. However, most environments of interest contain spatial patterns that are better characterized by models that capture dependencies among grid blocks. To account for such dependencies, we model the environment as a pairwise Markov random field. We specify a belief propagation-based mapping algorithm that takes these dependencies into account when estimating a map. To demonstrate the potential benefits of this approach, we simulate a simple multi-robot minefield mapping scenario. Minefields contain spatial dependencies since some landmine configurations are more likely than others, and since clutter, which causes false alarms, can be concentrated in certain regions and completely absent in others. Our belief propagation-based approach outperforms conventional occupancy grid mapping algorithms in the sense that better maps can be obtained with significantly fewer robot measurements. The belief propagation algorithm requires a modest amount of increased computation, but we contend that in applications where significant energy and time expenditure is associated with robot movement and active sensing, the reduction in the required number of samples justified the increased computation.

This paper computes the sensing capacity of a sensor network, with sensors of limited range, sensing a two-dimensional Markov random field, by modeling the sensing operation as an encoder. Sensor observations are dependent across sensors, and the sensor network output across different states of the environment is neither identically nor independently distributed. Using a random coding argument, based on the theory of types, we prove a lower bound on the sensing capacity of the network, which characterizes the ability of the sensor network to distinguish among environments with Markov structure, to within a desired accuracy

We propose a primitive, called Pioneer, as a first step towards verifiable code execution on untrusted legacy hosts. Pioneer does not require any hardware support such as secure co-processors or CPU-architecture extensions. We implement Pioneer on an Intel Pentium IV Xeon processor. Pioneer can be used as a basic building block to build security systems. We demonstrate this by building a kernel rootkit detector.

This paper develops a physical model of a MEMS capacitive accelerometer in order to use the accelerometer effectively in low-g motion tracking applications. The proposed physical model includes common physical parameters used to rate an accelerometer: scale factor, bias, and misalignment. Simple experiments used to reveal the behavior and characteristics of these parameters are described. A phenomenological modeling method is used to establish mathematical representations of these parameters in relation to errors such as nonlinearity, hysteresis, cross-axis effect, and temperature effect, without requiring a complete understanding of the underlying physics. Experimental results are presented, in which the physical model reduces RMSE by 93.1% in comparison with the manufacturer's recommended method.

We present a computational approach to learning by observation (LBO) that allows users to program mobile robots by demonstrating a task. Unlike previous approaches, our system incorporates statistical-learning techniques and concepts from control theory to reduce the amount of domain knowledge needed to infer the intent of the user. To improve the generalization ability of the system, the user can demonstrate the task multiple times. We extract task subgoals from these demonstrations and automatically associate them with objects in the environment. As these objects move, the subgoals are updated accordingly. This gives our system the ability to learn from demonstrations performed in different environments. In this paper, we present the concepts used in our LBO system as well as experimental laboratory results in learning motor-skill tasks.

In this paper we present a novel approach for representing trajectories using sequenced linear dynamical systems. This method uses a closed-form least-squares procedure to fit a single linear dynamical system (LDS) to a simple trajectory. These LDS estimates form the elemental building blocks used to describe complicated trajectories through an automatic segmentation procedure that can represent complicated trajectories with high accuracy. Each estimated LDS induces a control law, mapping current state to desired state, that encodes the target trajectory in a generative manner. We provide a proof of stability of the control law and show how multiple trajectories can be incorporated to improve the generalization ability of the system.

For a team of cooperating robots, geometry plays a vital role in operation. Knowledge of line of sight to local obstacles and adjacent teammates is critical in both the movement and planning stages to avoid collisions, maintain formation and localize the team. However, determining if other robots are within the line of sight of one another is difficult with existing sensor platforms - especially as the scale of the robot is reduced. We describe a method of exploiting collective team information to generate a virtual sensor that provides line of sight determination, greater range and resolution and the ability to generalize local sensing. We develop this sensor and apply it to the control of a tightly coupled, resource-limited robot team called Millibots.

This work describes a method for observing maneuvering targets using a group of mobile robots equipped with video cameras. These robots are part of a team of small-size (7/spl times/7/spl times/7 cm) robots configured from modular components that collaborate to accomplish a given task. The cameras seek to observe the target while facing it as much as possible from their respective viewpoints. This work considers the problem of scheduling and maneuvering the cameras based on the evaluation of their current positions in terms of how well can they maintain a frontal view of the target. We describe our approach, which distributes the task among several robots and avoids extensive energy consumption on a single robot. We explore the concept in simulation and present results.

This paper presents the theory and modeling of a quaternion-based augmented state Kalman filter for real-time orientation tracking of a handheld microsurgical instrument equipped with a magnetometer-aided all-accelerometer inertial measurement unit (IMU). The onboard sensing system provides two complementary sources of orientation information. The all-accelerometer IMU provides a high resolution but drifting angular velocity estimate, while the magnetic north vector is combined with the estimated gravity vector to yield a non-drifting but noisy orientation estimate. Analysis of the dominant stochastic noise components of the sensors and derivation of the noise covariance are presented. The proposed Kalman filter obtains a non-drifting orientation estimate with improved resolution by incorporating the motion dynamics of the instrument during microsurgery and models the angular velocity drift explicitly as extra dynamic states.

We define a notion of 'sensing capacity' that characterizes the ability of a sensor network to successfully distinguish among a discrete set of targets. Sensing capacity is defined as the maximum ratio of target positions to sensors for which inference of targets within a certain distortion is achievable. We demonstrate a lower bound on this capacity. Unlike previous work on 'sensor network capacity', our notion of sensing capacity is defined by the sensing task itself, as opposed to external resource constraints such as power, communications, and processing.

We expect a future where we are surrounded by embedded devices, ranging from Java-enabled cell phones to sensor networks and smart appliances. An adversary can compromise our privacy and safety by maliciously modifying the memory contents of these embedded devices. In this paper, we propose a softWare-based attestation technique (SWATT) to verify the memory contents of embedded devices and establish the absence of malicious changes to the memory contents. SWATT does not need physical access to the device's memory, yet provides memory content attestation similar to TCG or NGSCB without requiring secure hardware. SWATT can detect any change in memory contents with high probability, thus detecting viruses, unexpected configuration settings, and Trojan Horses. To circumvent SWATT, we expect that an attacker needs to change the hardware to hide memory content changes. We present an implementation of SWATT in off-the-shelf sensor network devices, which enables us to verify the contents of the program memory even while the sensor node is running.

In an ad-hoc network, intermediate nodes on a communication path are expected to forward packets of other nodes so that the mobile nodes can communicate beyond their wireless transmission range. However, because wireless mobile nodes are usually constrained by limited power and computation resources, a selfish node may be unwilling to spend its resources in forwarding packets which are not of its direct interest, even though it expects other nodes to forward its packets to the destination. It has been shown that the presence of such selfish nodes degrades the overall performance of a non-cooperative ad hoc network. To address this problem, we propose a secure and objective reputation-based incentive (SORI) scheme to encourage packet forwarding and discipline selfish behavior. Different from the existing schemes, under our approach, the reputation of a node is quantified by objective measures, and the propagation of reputation is efficiently secured by a one-way-hash-chain-based authentication scheme. Armed with the reputation-based mechanism, we design a punishment scheme to penalize selfish nodes. The experimental results show that the proposed scheme can successfully identify selfish nodes and punish them accordingly.

This paper presents a novel algorithm for synthesizing and editing video of natural phenomena that exhibit continuous flow patterns. The algorithm analyzes the motion of textured particles in the input video along user-specified flow lines, and synthesizes seamless video of arbitrary length by enforcing temporal continuity along a second set of user-specified flow lines. The algorithm is simple to implement and use. We used this technique to edit video of water-falls, rivers, flames, and smoke.

We present the design of an all-accelerometer inertial measurement unit (IMU). The IMU forms part of an intelligent hand-held microsurgical instrument that senses its own motion, distinguishes between hand tremor and intended motion, and compensates in real-time the erroneous motion. The new IMU design consists of three miniature dual-axis accelerometers, two of which are housed in a sensor suite at the distal end of the instrument handle, and one located at the proximal end close to the instrument tip. By taking the difference between the accelerometer readings, we decouple the inertial and gravitational accelerations from the rotation-induced (centripetal and tangential) accelerations, hence simplifies the kinematic computation of angular motions. We have shown that the error variance of the Euler orientation parameters /spl theta//sub x/, /spl theta//sub y/ and /spl theta//sub z/ is inversely proportional to the square of the distance between the three sensor locations. Comparing with a conventional three gyros and three accelerometers IMU, the proposed design reduces the standard deviation of the estimates of translational displacements by 29.3% in each principal axis and those of the Euler orientation parameters /spl theta//sub x/, /spl theta//sub y/ and /spl theta//sub z/ by 99.1%, 99.1% and 92.8% respectively.

Hysteresis of a piezoelectric actuator is rate dependent. Most hysteresis models are based on elementary rate independent operators and are not suitable for modeling actuator behavior across a wide frequency band. This work proposes a rate dependent modified Prandtl-Ishlinskii (PI) operator to account for the hysteresis of a piezoelectric actuator at varying frequency. We have shown experimentally that the relationship between the slope of the hysteretic loading curve and the rate of control input can be modeled by a linear function. The proposed rate-dependent hysteresis model is implemented for open-loop control of a piezoelectric actuator. In experiments tracking multi-frequency nonstationary motion profiles, it consistently outperforms its rate-independent counterpart by a factor of two in maximum error and a factor of three in rms error.

One of the main obstacles to automating production is the time needed to program the robot. Decreasing the programming time would increase the appeal of automation in many industries. In this paper we analyze the performance of a Predictive Robot Programming (PRP) system on complex, real-world robotic tasks. The PRP system attempts to decrease programming time by predicting the waypoints of a robot program based on previous examples of user behavior. We show that the PRP system is able to generate a large percentage of useful and highly accurate predictions, resulting in a potentially great amount of time saved.

We develop a new occupancy map that respects the role of the sensor measurement bearing and how it relates to the resolution of the existing occupancy map. We borrow an idea from Konolige for recording and tracking, in an occupancy-like map, the bearing at which sensor readings originate with respect to a given cell. Our specific contribution is in the way we process the sensor pose information, which is the bearing of the sensor readings when it indicates the presence of an obstacle in a particular cell. For each cell in the occupancy map, we calculate the greatest separation of incident poses, and then store that information in a new two-dimensional array called a pose map. A cell in the pose map measures the quality of information contained in the corresponding cell of the occupancy map. We merge the new pose map with the existing map to generate an enhanced occupancy map. Exploration plans derived from the enhanced occupancy map are more efficient and complete in that they do not guide the robot around phantom obstacles nor incorrectly classify narrow openings as closed commonly found in conventional occupancy maps.

We describe a new paradigm for exploration of unknown spaces based on maximizing the understanding of obstacles rather than the exposure of free space. We look at the interaction between multiple sensor readings and how they combine to resolve obstacles. Taking a next best view approach, we generate an inverse sensor model that identifies regions in space where a new sensor reading has maximal utility with respect to increasing the resolution of that reading. Fusion of multiple models is exploited to generate regions of interest that direct exploration in such a way as to maximize the robots understanding of its space. These techniques are applied to a team of small robots called Millibots.

As robots enter the human environment, there are increasing needs for novice users to be able to program robots with ease. A successful robot programming system should be intuitive, interactive, and intention aware. Intuitiveness refers to the use of intuitive user interfaces such as speech and hand gestures. Interactivity refers to the system's ability to let the user interact preemptively with the robot to take its control at any given time. Intention awareness refers to the system's ability to recognize and adapt to user intent. This paper focuses on the intention awareness problem for interactive multi-modal robot programming system. In our framework, user intent takes on the form of a robot program, which in our context is a sequential set of commands with parameters. To solve the intention recognition and adaptation problem, the system converts robot programs into a set of Markov chains. The system can then deduce the most likely program the user intends to execute based on a given observation sequence. It then adapts this program based on additional interaction. The system is implemented on a mobile vacuum cleaning robot with a user who is wearing sensor gloves, inductive position sensors, and a microphone.

Behavior-based design has been widely recognized as one of the main approaches in traditional robot design. In this paper, we propose an integration of behavior-based approach and evolutionary algorithm, called an evolutionary behavior programming system. By adapting from the idea of Behavior Analysis, Behavioral Modules and Interactions are used for coding behavior-based control systems into a particular programming form. Therefore, the processes of Genetic Programming (GP) can be performed to evolve possible behavior-based control systems. Furthermore, with the intention to improve the learning performances in dynamic environments, a new idea of turning on/off each module in the network stochastically, called Dynamic Network, is applied. Experiments on soccer robots are simulated to verify this new approach. With the intention of training robots to play soccer games, staged evolution is used to speed up the learning process. Simulation experimental results demonstrate that this proposed method is highly promising.

Deployed vision systems often encounter image variations poorly represented in their training data. While observing their environment, such vision systems obtain unlabeled data that could be used to compensate for incomplete training. In order to exploit these relatively cheap and abundant unlabeled data we present a family of algorithms called /spl lambda/MEEM. Using these algorithms, we train an appearance-based people detection model. In contrast to approaches that rely on a large number of manually labeled training points, we use a partially labeled data set to capture appearance variation. One can both avoid the tedium of additional manual labeling and obtain improved detection performance by augmenting a labeled training set with unlabeled data. Further, enlarging the original training set with new unlabeled points enables the update of detection models after deployment without human intervention. To support these claim we show people detection results, and compare our performance to a purely generative expectation maximization-based approach to learning over partially labeled data.

The effectiveness of multirobot learning in achieving optimal, cooperative solutions is potentially affected by various factors having to do with the nature and configuration of the robots and the nature and configuration of the robots and the nature of the learning entities. Varying one factor wrongly may lead to undesirable results. There is no reported work on how systematically to set up these factors. In this paper, we methodically test the effect of varying four common factors (reward scope, global information delay, diversity of robots, and number of robots) in a decentralized multirobot system, first in simulation and then on real robots. The results show that two of these factors, reward scope and global information delay, if set up incorrectly, can prevent optimal, cooperative solutions.

The invention of the Jacquard weaving machine led to the concept of a stored "program" and "mechanized" binary information processing. This development served as the inspiration for C. Babbage's analytical engine-the precursor to the modern-day computer. Today, more than 200 years later, the link between textiles and computing is more realistic than ever. In this paper, we look at the synergistic relationship between textiles and computing and identify the need for their "integration" using tools provided by an emerging new field of research that combines the strengths and capabilities of electronics and textiles into one: electronic textiles, or e-textiles. E-textiles, also called smart fabrics, have not only "wearable" capabilities like any other garment, but also have local monitoring and computation, as well as wireless communication capabilities. Sensors and simple computational elements are embedded in e-textiles, as well as built into yarns, with the goal of gathering sensitive information, monitoring vital statistics, and sending them remotely (possibly over a wireless channel) for further processing. The paper provides an overview of existing efforts and associated challenges in this area, while describing possible venues and opportunities for future research.

Scaling in CMOS device technology has made it possible to cheaply embed intelligence in a myriad of devices. In particular, it has become feasible to fabricate flexible materials (e.g., woven fabrics) with large numbers of computing and communication elements embedded into them. Such computational fabrics, electronic textiles, or e-textiles have applications ranging from smart materials for aerospace applications to wearable computing. This paper addresses the modeling of computation, communication and failure in e-textiles and investigates the performance of two techniques, code migration and remote execution, for adapting applications executing over the hardware substrate, to failures in both devices and interconnection links. The investigation is carried out using a cycle-accurate simulation environment developed to model computation, power consumption, and node/link failures for large numbers of computing elements in configurable network topologies. A detailed analysis of the two techniques for adapting applications to the error prone substrate is presented, as well as a study of the effects of parameters, such as failure rates, communication speeds, and topologies, on the efficacy of the techniques and the performance of the system as a whole. It is shown that code migration and remote execution provide feasible methods for adapting applications to take advantage of redundancy in the presence of failures and involve trade offs in communication versus memory requirements in processing elements.

This paper describes research in active instruments for enhanced accuracy in microsurgery. The aim is to make accuracy enhancement as transparent to the surgeon as possible. Rather than using a robotic arm, we have taken the novel approach of developing a handheld instrument that senses its own movement, distinguishes between desired and undesired motion, and deflects its tip to perform active compensation of the undesired component. The research has therefore required work in quantification and modeling of instrument motion, filtering algorithms for tremor and other erroneous movements, and development of handheld electromechanical systems to perform active error compensation. The paper introduces the systems developed in this research and presents preliminary results.

This paper addresses an emerging new field of research that combines the strengths and capabilities of electronics and textiles in one: electronic textiles, or e-textiles. E-textiles, also called Smart Fabrics, have not only "wearable" capabilities like any other garment, but also local monitoring and computation, as well as wireless communication capabilities. Sensors and simple computational elements are embedded in e-textiles, as well as built into yarns, with the goal of gathering sensitive information, monitoring vital statistics and sending them remotely (possibly over a wireless channel) for further processing. Possible applications include medical (infant or patient) monitoring, personal information processing systems, or remote monitoring of deployed personnel in military or space applications. We illustrate the challenges imposed by the dual textile/electronics technology on their modeling and optimization methodology.

To date, very little work has investigated the benefits of repairable robots. Robots that can repair themselves and other robots in their team are intuitively a superior design. Intuition, however, is not an acceptable basis for spending millions of dollars in development. In this work, we quantify the gain in productivity of a team of repairable robots compared to a team without repair capabilities. We create a model using an extension of standard reliability theory. It allows the definition of a metric which is used to compare the two teams. The analysis yields insight into the design of repairable robot teams under a certain set of assumptions. The model also demonstrates scenarios where repair capabilities are not likely to be beneficial.

This paper introduces a novel approach to program a robot interactively through a multi-modal interface. The key characteristic of this approach is that the user can provide feedback interactively at any time - during both the programming and the execution phase. The framework takes a three-step approach to the problem: multi-modal recognition, intention interpretation, and prioritized task execution. The multi-modal recognition module translates hand gestures and spontaneous speech into a structured symbolic data stream without abstracting away the user's intent. The intention interpretation module selects the appropriate primitives to generate a task based on the user's input, the system's current state, and robot sensor data. Finally, the prioritized task execution module selects and executes skill primitives based on the system's current state, sensor inputs, and prior tasks. The framework is demonstrated by interactively controlling and programming a vacuum-cleaning robot.

Describes a multi-layered hierarchical motion planning strategy for a class of self-reconfigurable modular robotic systems, I-Cubes. The approach is based on the synthesis of motion on the basis of metacubes, which have a particular structure possessing 8 Cubes and 16 Links. The developed strategy organizes the metacube motions and the corresponding cube-level motions. At the lowest level, link motions are generated. The resulting system is demonstrated to be capable of performing a pre-specified task of moving from one position/shape to another. The paper describes the latest results of our planning strategy through some experimentally justified examples.

Learning can be an effective way for robot systems to deal with dynamic environments and changing task conditions. However, popular single-robot learning algorithms based on discounted rewards, such as Q learning, do not achieve cooperation (i.e., purposeful division of labor) when applied to task-level multirobot systems. A task-level system is defined as one performing a mission that is decomposed into subtasks shared among robots. We demonstrate the superiority of average-reward-based learning such as the Monte Carlo algorithm for task-level multirobot systems, and suggest an explanation for this superiority.

One of the main barriers to automating a particular task with a robot is the amount of time needed to program the robot. Decreasing the programming time would facilitate automation in domains previously off limits. In this paper, we present a novel method for leveraging the previous work of a user to decrease future programming time: predictive robot programming. The decrease in programming time is accomplished by predicting waypoints in future robot programs and automatically moving the manipulator end-effector to the predicted position. To this end, we develop algorithms that construct simple continuous-density hidden Markov models by a state-merging algorithm based on waypoints from prior robot programs. We then use these models to predict the waypoints in future robot programs. While the focus of this paper is the application of predictive robot programming, we also give an overview of the underlying algorithms used and present experimental results.

A behavior-based approach has been effectively applied for the design of robot control systems, and evolutionary algorithms have been implemented as an approach to generate the robot control systems automatically. In this paper, we propose the integration of both concepts as an automatic behavior programming system. By adapting the idea of behavior analysis, behavioral modules and interactions are presented in order to be able to represent behavior-based control systems in a programming paradigm. Then, by manipulating the program codes without human intervention, the processes of Genetic Programming (GP) are applied to discover the possible behavior-based control systems, which successfully solve the given problems. Moreover, with the intention of improving the learning performance in dynamic environments, the new idea of turning on/off each node in the network stochastically, called a Dynamic Network (DN), is applied. Experimental results show the potential of our approach.

The objective of the CyberScout project is to develop an autonomous surveillance and reconnaissance system using a network of all-terrain vehicles. We focus on two facets of this system: 1) vision for surveillance and 2) autonomous navigation and dynamic path planning. In the area of vision-based surveillance, we have developed robust, efficient algorithms to detect, classify, and track moving objects of interest (person, people, or vehicle) with a static camera. Adaptation through feedback from the classifier and tracker allow the detector to use grayscale imagery, but perform as well as prior color-based detectors. We have extended the detector using scene mosaicing to detect and index moving objects when the camera is panning or tilting. The classification algorithm performs well with coarse inputs, has unparalleled rejection capabilities, and can flag novel moving objects. The tracking algorithm achieves highly accurate (96%) frame-to-frame correspondence for multiple moving objects in cluttered scenes by determining the discriminant relevance of object features. We have also developed a novel mission coordination architecture, CPAD (Checkpoint/Priority/Action Database), which performs path planning via checkpoint and dynamic priority assignment, using statistical estimates of the environment's motion structure. The motion structure is used to make both preplanning and reactive behaviors more efficient by applying global context. This approach is more computationally efficient than centralized approaches and exploits robot cooperation in dynamic environments better than decoupled approaches.

In the future, we propose that there will be largely self-sufficient robot colonies operating on distant planets and in harsh environments here on Earth. A highly desirable quality of such a colony would be the capability of the robots to repair each other. Towards the goal of autonomous repair, we design a robot that can replace the modules composing a similar robot. The final system is teleoperated and module removal/replacement is performed on a test bed. We discuss some of the design trade-offs for such a system and discuss some of the steps required in order to develop a self-sufficient robot colony.

A framework, called skill advice guided exploration (SAGE), for the adaptive transfer of robot skill knowledge using reinforcement learning (RL) agents is presented. A skill is viewed as a reactive policy which maps world states to agent actions. It may be acquired via learning or it may be hand-coded by the designer. The SAGE framework allows multiple, possibly conflicting, sources of knowledge to be incorporated simultaneously. An abstraction for knowledge in an RL system, called advice, is introduced. The advice abstraction permits the transfer of information between RL agents with differing internal representations. A SAGE-based system can learn to disregard misleading advice. The potential of this methodology is demonstrated on a set of discrete learning tasks. Results show that SAGE-based systems can benefit from relevant information and that incorrect information does not prevent learning of the task solution. The benefits, limitations, and possible extensions of this work are discussed.

Presents the development and. initial experimental results of the first prototype of Micron, an active hand-held instrument to sense and compensate physiological tremor and other unwanted movement during vitreoretinal microsurgery. The instrument incorporates six inertial sensors, allowing the motion of the tip to be computed. The motion captured is processed to discriminate between desired and undesired components of motion. Tremor canceling is implemented via the weighted-frequency Fourier linear combiner (WFLC) algorithm, and compensation of non-tremorous error via a neural network technique is being investigated. The instrument tip is attached to a three-degree-of-freedom parallel manipulator with piezoelectric actuation. The actuators move the tool tip in opposition to the tremor, thereby suppressing the erroneous motion. Motion canceling experiments with oscillatory motions in the frequency band of physiological tremor show that Micron is able to reduce error amplitude by 45.3% in 1-D tests and 37.2% in 3-D tests.

Knowledge of position in the context of its surrounding is necessary for robots to build maps and develop path plans. Limitations in odometry and the lack of a priori knowledge reduce the effectiveness of a single robot to retain a sense of position for any extended duration. The problem is only compounded when the scale of the robot is reduced. However, by employing multiple robots we can exploit their distributed nature to provide an external context in which to evaluate sensor readings for mapping and localization. We have designed a team of centimeter-sized robots that coordinate sensing and action to establish and maintain position as they move throughout space. By utilizing low-cost ultrasonic sensors, the team is able to measure the range between each robot pair. We pose these measurements in terms of a position likelihood and combine them to find a global solution that best maximizes the position likelihood of each robot. We also address a unique multipath interference mode that arises as a direct result of the reduced scale of the robot team. We present our experiences with localization and control of a small robot team.

In this paper, we present a multilayered planner for the motion of modules in a uniform group of I-Cube self-reconfiguring modular robotic system. The planner uses metacubes, the high-level abstraction of 8C16L groups. It combines distributed approaches at the high-level with lowlevel trajectory computation for the actual modules which can be completed in O(n) steps where n is the number of the cubes in the system, and pre-defined rules for link motions. Mechatronic properties of the latest version of the reconfiguring modules are presented as well as results of the latest hardware and software experiments with the I-Cube modules.

We propose a novel method for temporally and spatially corresponding moving objects by automatically learning the relevance of the objects' appearance features to the task of discrimination. Efficient correspondence is achieved by enforcing temporal consistency of the relevances for a particular object. Relevances are learned using a technique we have termed "differential discriminative diagnosis". An agent is assigned to each moving object in the scene. The agent possesses the basic capability to decide whether or not an object in the scene is the one it represents. Each agent customizes itself to the object by means of differential discriminative diagnosis as the object persists in the scene. We explain this correspondence scheme as applied to the task of corresponding moving people in a surveillance system.

Design and implementation of I-Cubes, a modular self-reconfigurable robotic system, is discussed. I-Cubes is a bipartite collection of individual modules that can be independently controlled. The group consists of active elements, called links, which are 3-DOF manipulators capable of attaching to/detaching from the passive elements (cubes) acting as connectors. The cubes can be oriented and positioned by the links. Using actuation and attachment properties of the link and the cubes, the system can self-reconfigure to adapt to its environment. Tasks such as moving over obstacles, climbing stairs can be performed by changing the relative position and connection of the modules. The links are actuated using servomotors and worm gear mechanisms. Mechanical encoders and rotary switches provide position feedback for semi-autonomous control of the system. The cubes are equipped with a novel mechanism that provides inter-module attachment. Design and hardware implementation of the system as well as experimental results are presented.

Self-adaptive control software is a new paradigm to create robust, fault-tolerant mobile robots. This type of software analyzes its performance and dynamically modifies itself to operate better in adverse and rapidly changing conditions. We have created two systems that facilitate the creation of self-adaptive control software: PB3A and RAVE. PB3A, the Port-Based Adaptable Agent Architecture, is a mobile, agent-based framework that allows software to adapt itself at all levels. RAVE, the Real And Virtual Environment, is a mixed-reality simulation environment for mobile robots. Together these two systems allow for the creation, testing, and analysis of self-adaptive control software by on- and off-line simulation. We give brief overviews of PB3A and RAVE and present applications that demonstrate robotic systems using self-adaptive control software.

In this paper we present an approach to teaching an introductory course in electrical and computer engineering to freshmen engineering students. The primary distinguishing characteristics of our approach include: providing information in an experiential context or "teaching-in-context"; focusing on "black-box"-based hierarchical decomposition as an approach to understanding complex systems; and using piecewise linear models for nonlinear elements as a way to develop intuition and understanding about the workings of circuits. The experiential context is provided by having students carry out a series of laboratory exercises involving an exciting real-world electronic system. To date, we have developed laboratory exercises around a programmable robot and a Global Positioning System receiver. The greatest challenges in teaching an intellectually substantive course to entering freshmen are their lack, on average, of the mathematics and physics background that is taken for granted in most electrical and computer engineering courses. In this paper we demonstrate that by appropriate choice of material and presentation methods this challenge can be overcome.

Presents a distributed planning and control architecture for autonomous multi-manipulator systems (MMS). The control architecture is implemented using an agent-based approach. A team of distributed and autonomous agents is deployed to model the flexible assembly system in such a way that the agents negotiate, collaborate, and cooperate to achieve the goals of assembly tasks. The main focus of the paper is on assembly task allocation and assembly task execution. We describe the agent models and communication mechanism, and explain how they handle complex interactions among agents. A distributed trajectory planning approach based on artificial potential fields is also presented. Experimental results show that our multi-agent planning and control framework is suitable for flexible robotic assembly tasks. Our approach addresses the issues of flexibility, scalability, reconfigurability, and fault-tolerance. We anticipate that the same approach can be applied to other flexible manufacturing environments.

Explores gesture-based programming as a paradigm for programming robotic agents. Gesture-based programming is a form of programming by human demonstration that focuses the development of robotic systems on task experts rather than programming experts. The technique relies on the existence of previously acquired robotic skills (which we call "sensorimotor primitives") which we hope to develop to into the robotic equivalent of skills acquired by humans through everyday experiences. The interpretation of the human's demonstration and subsequent matching to robotic primitives is a qualitative problem that we approach with a community of skilled agents. A simple manipulative task and a variant of that task are programmed to demonstrate the system.

To focus on the research issues surrounding collaborative behavior in multiple mobile-robotic systems, a great amount of low-level infrastructure is required. To facilitate our on-going research into multi-robot systems, we have developed RAVE, a software framework that provides a real and virtual environment for running and managing multiple heterogeneous mobile-robot systems. This framework simplifies the implementation and development of collaborative robotic systems by providing the following capabilities: the ability to run systems off-line in simulation, user-interfaces for observing and commanding simulated and real robots, transparent transference of simulated robot programs to real robots, the ability to have simulated robots interact with real robots, and the ability to place virtual sensors on real robots to augment or experiment with their performance.

Gestures provide a rich and intuitive form of interaction for controlling robots. This paper presents an approach for controlling a mobile robot with hand gestures. The system uses hidden Markov models (HMMs) to spot and recognize gestures captured with a data glove. To spot gestures from a sequence of hand positions that may include nongestures, we have introduced a "wait state" in the HMM. The system is currently capable of spotting six gestures reliably. These gestures are mapped to robot commands under two different modes of operation: local and global control. In the local control mode, the gestures are interpreted in the robot's local frame of reference, allowing the user to accelerate, decelerate, and turn. In the global control mode, the gestures are interpreted in the world frame, allowing the robot to move to the location at which the user is pointing.

Transforming software requirements into a software design involves the iterative partition of a solution into software components. The process is human-intensive and does not guarantee that design objectives such as reusability, evolvability, and reliable performance are satisfied. The costly process of designing, building, and modifying high-assurance systems motivates the need for precise methods and tools to generate designs whose corresponding implementations are reusable, evolvable, and reliable. This paper demonstrates an analytical approach for partitioning basic elements of a software solution into reusable and evolvable software components. First, we briefly overview the role of partitioning in current design methods and explain why computer-aided design (CAD) tools to automate the design of microelectromechanical systems (MEMS) are high-assurance applications. Then we present our approach and apply it to the design of CAD software to layout an optimized design of a MEMS accelerometer to be used in the navigational units of aircraft. Lastly, we discuss the implications of our approach and future research directions.

A manipulation task primitive is classified by the relative motion between two (rigid) parts. Only twenty different relative motions are possible and these can be used to guide the identification and development of manipulation task primitives. The goal is to build a richer library of robot capabilities in the manipulation domain. By identifying manipulation task primitives and instantiating solutions to them with available sensors and robot hardware in the form of sensorimotor primitives, we provide a higher-level abstraction for composing solutions to complex manipulation tasks. A key benefit is the ability to re-use costly sensor-based control algorithms for executing these primitives. We discuss the implementation of a few manipulation task primitives using force damping control and active vision feedback. Finally, we decompose a common task into two different skills using the primitives described.

Visual servoing is a robust technique for aligning both static and moving parts using imprecisely calibrated camera-lens-manipulator systems. An important limitation of these systems is the workspace within which the alignment task can be successfully performed due to the position and orientation of the camera. An active camera can extend this region, however this changes the visual representation of the task itself. Therefore, the reference input that drives the visually servoed manipulator must change appropriately. In this paper a framework that allows for camera-lens motion during visually servoed manipulation is described. The main components of the framework include object schemas and port-based agents. Object schemas represent the task internally in terms of geometric models with attached sensor mappings. Object schemas are dynamically updated by sensor feedback, and thus provide an ability to perform three dimensional spatial reasoning during task execution, a capability traditional image-based visual servoing lacks. Object schemas are also able to dynamically create desired visual representations of the task from which reference inputs for vision-based control strategies are derived. The sensor mappings of object schemas are also used to guide camera motion based on task characteristics. Port-based agents are the executors of the visual reference inputs and the camera motion commands. They interact with the real world through visual servoing control laws. Experimental results that demonstrate system capabilities and performance are presented.

Traditional telerobotic systems often provide feedback to the user through a variety of sensing modalities, for example through live video imagery, force reflection, or acoustic signals. When a supervisor guides a robotic task while immersed in a virtual environment, geometric representations of the world are provided as feedback. In this case, the supervisor interacts with the environment by visually observing these virtual objects and directing their motion. At issue is how to appropriately use various sensing modalities provided by disparate sensors in a system of this configuration. This paper focuses on assimilating the disparate feedback provided by force and vision sensors for telerobotic systems guided from within virtual environments. A framework for feedback assimilation is described based on the concept of sensor resolvability. Sensor resolvability is used in two ways, to update the virtual environment and to guide the desired task in the real world. Resolvability selects the appropriate sensing modality to use in updating the virtual environment within which the supervisor is immersed. It is also used to direct semi-autonomous agents that interact directly with the real world to perform the desired task. Experimental results demonstrate the significant advantages of assimilating disparate sensory feedback throughout a telerobotic task using the concept of sensor resolvability.

A rapidly deployable fault tolerant manipulator system consists of modular hardware and support software that allow the user to quickly configure and deploy a fault tolerant manipulator that is custom-tailored for a given task. The main focus of this paper is on the task based design component of such a system, that is, the determination of the optimal manipulator configuration, its base position and the corresponding joint space trajectory for a given task. We introduce a novel agent-based solution approach to task based design and illustrate it with a fault tolerant manipulator design for a satellite docking operation aboard the space shuttle.

We present a novel application of a neural network to augment manual precision by cancelling involuntary motion. This method may be applied in microsurgery, using either a telerobotic approach or active compensation in a handheld instrument. A feedforward neural network is trained to input the measured trajectory of a handheld tool tip and output the intended trajectory. Use of the neural network decreases rms error in recordings from four subjects by an average of 43.9%.

The tropism system cognitive architecture provides an intuitive formalism for colonies of agents, either hardware or software. We present a fine-grained implementation of the architecture on a colony of software agents for the interpretation of human tactile gestures for robotic trajectory specification and modification. The fine-grained nature of the architecture and the use of the port-based object framework for agent instantiation allows the manual construction of a capable agent set that is reconfigurable and reusable across different gesture-based interaction tasks.

In this paper we summarize recent research results from a new technique of sensor calibration called shape-from-motion calibration. We first present the basic technique, which is an eigenspace analysis, and show that it includes the rigor of least squares without the full burden of measuring all the applied inputs. Next we present new research that removes another constraint of the calibration technique and extends the robustness to cover systems with slight nonlinearities.

The port-based object is a new software abstraction for designing and implementing dynamically reconfigurable real-time software. It forms the basis of a programming model that uses domain-specific elemental units to provide specific, yet flexible, guidelines to control engineers for creating and integrating software components. We use a port-based object abstraction, based on combining the notion of an object with the port-automaton algebraic model of concurrent processes. It is supported by an implementation using domain-specific communication mechanisms and templates that have been incorporated into the Chimera real-time operating system and applied to several robotic applications. This paper describes the port-based object abstraction, provides a detailed analysis of communication and synchronization based on distributed shared memory, and describes a programming paradigm based on a framework process and code templates for quickly implementing applications.

This paper describes a multi-agent approach to grasping that combines collision avoidance and visual servoing strategies. Control of the system is distributed among five different types of agents: link agents, joint agents, end-effector agents, task agents, and object agents. As each agent attempts to achieve a desired individual behavior, the manipulator itself exhibits an emergent behavior that avoids obstacles while approaching an object to be grasped. As the object is approached, object agents that represent the object to be grasped and the gripper performing the grasp are instantiated. The instantiation of each of these two object agents, which are guided by visual feedback, automatically changes the manipulator control structure. Once both object agents become active, closed-loop visual servoing of the end-effector during the final stage of grasping is realized. Experimental results are presented that illustrate the three types of manipulator control realized during a typical grasping task.

A rapidly deployable manipulator system combines the flexibility of reconfigurable modular hardware with modular programming tools, allowing the user to rapidly create a manipulator which is custom-tailored for a given task. This article describes two main aspects of such a system, namely, the reconfigurable modular manipulator system (RMMS) hardware and the corresponding control software.

We present the design of a modular tactile sensor and actuator system for observing human demonstrations of contact tasks. The system consists of three interchangeable parts: an intrinsic tactile sensor for measuring net force/torque, an extrinsic tactile sensor for measuring contact distributions, and a tactile actuator for displaying tactile distributions. The novel components are the extrinsic sensor and tactile actuator which are "inside-out symmetric" to each other and employ an electrorheological gel for actuation.

The Chimera Methodology is a software engineering paradigm that enables rapid development of real-time applications through use of dynamically reconfigurable and reusable software. It is targeted towards a distributed shared memory computing environment. The primary contribution of this research is the port-based object model of a real-time software component. The model is obtained by applying the port-automaton formal computational theory to object-based design. A finite state machine, detailed interface specifications, and a C-language template are used to define the port-based object. Tools to support the in-tegration, scheduling, and state variable communication between the objects have been developed and incorporated into the Chimera Real-Time Operating System. Techniques for verifying correctness and analyzing performance are also provided for configuration managers that integrate software designed using the port-based object model.

In this paper, we include Coulomb friction at contacts between bodies and give a characterization of the entire set of stable orientations of an assembly under uniform gravity. Our characterization is based on the concept of potential stability, which describes a necessary but not sufficient condition for the stability of an assembly. Orientations that are computed as being unstable, however, are guaranteed to fall apart. Our characterization reveals that the set of stable orientations maps out a convex region on the unit-sphere of directions and corresponds to a spherical analog of a planar polygon-the region is bounded by a sequence of vertices joined by great arcs. Linear programming techniques are used to automatically find this set of vertices, yielding a description of the range of stable orientations for any assembly. For frictionless assemblies, our characterization of stable orientations is exact. For assemblies with friction, some conservative approximations associated with the use of a linearized Coulomb law are made.

Force and vision sensors provide complementary information, yet they are fundamentally different sensing modalities. This implies that traditional sensor integration techniques that require common data representations are not appropriate for combining the feedback from these two disparate sensors. In this paper, we introduce the concept of vision and force sensor resolvability as a means of comparing the ability of the two sensing modes to provide useful information during robotic manipulation tasks. By monitoring the resolvability of the two sensing modes with respect to the task, the information provided by the disparate sensors can be seamlessly assimilated during task execution. A nonlinear force/vision servoing algorithm that uses force and vision resolvability to switch manipulator control between sensors is proposed. The advantages of the assimilation technique are demonstrated during contact transitions between a stiff manipulator and a rigid environment, a system configuration that easily becomes unstable when force control alone is used. Experimental results show that robust contact transitions are made by the proposed nonlinear controller while simultaneously satisfying the conflicting task requirements of fast approach velocities, maintaining stability, minimizing impact forces, and suppressing bounce between contact surfaces.

Many researchers are interested in developing robust, skill-achieving robot programs. The authors propose the development of a sensorimotor primitive layer which bridges the gap between the robot/sensor system and a class of tasks by providing useful encapsulations of sensing and action. Skills can then be constructed from this library of sensor-driven primitives. This reflects a move away from the separation of sensing and action in robot programming of task strategies towards the integration of sensing and action in a domain-general way for broad classes of tasks. For the domain of rigid-body assembly, the authors exploit the motion constraints which define assembly to develop force sensor-driven primitives. The authors report on the experimental results of a D-connector insertion skill implemented using several force-driven primitives.

Visual servoing is a manipulation control strategy that precisely positions objects using imprecisely calibrated camera-lens-manipulator systems. In order to quickly and easily integrate sensor-based manipulation strategies such as visual servoing into robotic systems, a system framework and a task representation must exist which facilitates this integration. The framework must also be extendable so that obsolete sensor systems can be easily replaced or extended as new technologies become available. In this paper we present a framework for expectation-based visual servoing which visually guides tasks based on the expected visual appearance of the task. The appearance of the task is generated by a model of the environment that uses texture-mapped geometric models to represent objects. A system structure which facilitates the integration of various configurations of visual servoing systems is presented, as well as a hardware implementation of the proposed system and experimental results using a stereo camera system.

Integrating sensors into robot systems is an important step towards increasing the flexibility of robotic manufacturing systems. Current sensor integration is largely task-specific which hinders flexibility. The authors are developing a sensorimotor command layer that encapsulates useful combinations of sensing and action which can be applied to many tasks within a domain. The sensorimotor commands provide a higher-level in which to terminate task strategy plans, which eases the development of sensor-driven robot programs. This paper reports on the development of both force and vision driven commands which are successfully applied to two different connector insertion experiments.

Ideal manipulator end-effector transitions from noncontact to contact states should be fast and stable with minimal impact forces and without bounce. These specifications, however are difficult to simultaneously achieve, especially for the most common manipulator force control configuration employing a wrist force sensor and a stiff manipulator position loop. In this paper the authors present a control strategy that uses high bandwidth vision feedback (30 Hz) in addition to force feedback (100 Hz) for contact transient control. A nonlinear control strategy is proposed that considers force and vision feedback simultaneously and then switches to pure force control when the camera-lens system becomes unable to accurately resolve the location of the end-effector relative to the surface to be contacted. Experimental results are presented which demonstrate that a stiff manipulator can quickly contact a stiff surface stably without bounce. Results also show that the magnitude of the impact force spike is directly related to the accuracy with which contact surfaces are visually observed.

Whether a task can be completed after a failure of one of the degrees-of-freedom of a redundant manipulator depends on the joint angle at which the failure takes place. It is possible to achieve fault tolerance by globally planning a trajectory that avoids unfavorable joint positions before a failure occurs. In this paper we present a trajectory planning algorithm that guarantees fault tolerance while simultaneously satisfying joint limit and obstacle avoidance requirements.

Component-based real-time software speeds development and lowers cost of robotics applications. It enables the use of rapid prototyping or incremental software process models. The Chimera Methodology is a software engineering paradigm targeted at developing and integrating dynamically reconfigurable and reusable real-time software components. It is founded upon the notion of port-based objects. The focus of this paper is how to apply the Chimera Methodology specifically to the development of robotic applications.

Gesture-based programming is a new paradigm to ease the burden of programming robots. By tapping in to the user's wealth of experience with contact transitions, compliance, uncertainty and operations sequencing, we hope to provide a more intuitive programming environment for complex, real-world tasks based on the expressiveness of nonverbal communication. A requirement for this to be accomplished is the ability to interpret gestures to infer the intentions behind them. As a first step toward this goal, this paper presents an application of distributed perception for inferring a user's intentions by observing tactile gestures. These gestures consist of sparse, inexact, physical "nudges" applied to the robot's end effector for the purpose of modifying its trajectory in free space. A set of independent agents-each with its own local, fuzzified, heuristic model of a particular trajectory parameter observes data from a wristforce/torque sensor to evaluate the gestures. The agents then independently determine the confidence of their respective findings and distributed arbitration resolves the interpretation through voting.

In the Fall of 1991, after approximately two years of development, the department of Electrical and Computer Engineering (ECE) at Carnegie Mellon University (CMU) implemented a new curriculum that differed radically from its predecessor. Key features of this curriculum include: Engineering in the Freshman year, a small core of required classes, area requirements in place of most specific course requirements, mandated breadth, depth, design, and coverage across ECE technical areas, a relatively large fraction of free electives, and a single integrated Bachelor of Science degree in Electrical and Computer Engineering. In this paper we review the design of this curriculum, including a taxonomy of problems we needed to address, and a set of general principles we evolved to address them. The new curriculum is described in detail, including new data from an ongoing analysis of its impact on students' curricula choices.>

The stability of an assemblage of contacting rigid bodies without friction is investigated. A method is presented for finding an orientation of the assembly so that the assembly remains motionless under gravity. If no stable orientation exists for an assembly, the method finds the "least" unstable orientation. The metric used to measure stability is based on the second time-rate of change of the gravitational potential energy, and the desired orientation for an assembly is expressed in terms of an optimization problem involving changes in potential energy. The problem of finding stable or maximally-stable orientations is formulated as a constrained maximin problem. The maximin problem is shown to be a variant of standard zero-sum matrix games, and can be solved using linear programming. The method is the first general method for automatically determining stable orientations. Example assemblies are presented.>

Algorithms for full 3D robotic visual tracking of moving targets whose motion is 3D and consists of translational and rotational component are presented. The objective of the system is to track selected features on moving objects and to place their projections on the image plane at desired positions by appropriate camera motion. The most important characteristics of the proposed algorithms are the use of a single camera mounted on the end-effector of a robotic manipulator (eye-in-hand configuration), and the fact that these algorithms do not require accurate knowledge of the relative distance of the target object from the camera frame. The detection of motion is based on a cross-correlation technique known as sum-of squares differences (SSD) algorithm. The camera model used introduces a number of parameters that are estimated online, further reducing the algorithms' reliance on precise calibration of the system. An adaptive control algorithm compensates for modeling errors, tracking errors, and unavoidable computations delays which result from time-consuming image processing.>

This paper introduces a sensor placement measure called resolvability. The measure provides a technique for estimating the relative ability of various visual sensors, including monocular systems, stereo pairs, multi-baseline stereo systems, and 3D rangefinders, to accurately control visually manipulated objects. The resolvability ellipsoid illustrates the directional nature of resolvability, and can be used to direct camera motion and adjust camera intrinsic parameters in real-time so that the servoing accuracy of the visual servoing system improves with camera-lens motion. The Jacobian mapping from task space to sensor space is derived for a monocular system, a stereo pair with parallel optical axes, and a stereo pair with perpendicular optical axes. Resolvability ellipsoids based on these mappings for various sensor configurations are presented. Visual servoing experiments demonstrate that resolvability can be used to direct camera-lens motion in order to increase the ability of a visually servoed manipulator to precisely servo objects.>

Previous work by Mattikalli et al. (1993) considered the stability of assemblies of frictionless contacting bodies with uniform gravity. A linear programming-based technique was described that would automatically determine a single stable orientation for an assembly (if such an orientation existed). In this paper, we give an exact characterization of the entire set of stable orientations of any assembly under uniform gravity. Our characterization reveals that the set of stable orientations maps out a convex region on the unit-sphere of directions. The region is bounded by a sequence of vertices adjoined with great arcs. Linear programming techniques are used to automatically find this set of vertices, yielding a precise description of the range of stable orientations for any frictionless assembly.>

Real-time visual feedback is an important capability that many robotic systems must possess if these systems are to operate successfully in dynamically varying and imprecisely calibrated environments. An eye-in-hand system is a common technique for providing camera motion to increase the working region of a visual sensor. Although eye-in-hand robotic systems have been well-studied, several deficiencies in proposed systems make them inadequate for actual use. Typically, the systems fail if manipulators pass through singularities or joint limits. Objects being tracked can be lost if the objects become defocused, occluded, or if features on the objects lie outside the field of view of the camera. In this paper, a technique is introduced for integrating a visual tracking strategy with dynamically determined sensor placement criteria. This allows the system to automatically determine, in real-time, proper camera motion for tracking objects successfully while accounting for the undesirable, but often unavoidable, characteristics of camera-lens and manipulator systems. The sensor placement criteria considered include focus, field-of-view spatial resolution, manipulator configuration, and a newly introduced measure called resolvability. Experimental results are presented.>

Algorithms for 3D robotic visual tracking of moving targets whose motion is 3D and consists of translational and rotational components are presented. The objective of the system is to track selected features on moving objects and to place their projections on the image plane at desired positions by appropriate camera motion. The most important characteristics of the proposed algorithms are the use of a single camera mounted on the end-effector of a robotic manipulator (eye-in-hand configuration), and the fact that these algorithms do not require accurate knowledge of the relative distance of the target object from the camera frame. This fact makes these algorithms particularly useful in environments that are difficult to calibrate. The camera model used introduces a number of parameters that are estimated on-line, further reducing the algorithms' reliance on precise calibration of the system. An adaptive control algorithm compensates for modeling errors, tracking errors, and unavoidable computational delays which result from time-consuming image processing. Experimental results are presented to verify the efficacy of the proposed algorithms. These experiments were performed using a multi-robotic system consisting of Puma 560 manipulators.>

The application of robots in critical missions in hazardous environments requires the development of reliable or fault tolerant manipulators. In this paper we define fault tolerance as the ability to continue the performance of a task after immobilization of a joint due to failure. Initially, no joint limits are considered, in which case we prove the existence of fault tolerant manipulators and develop an analysis tool to determine the fault tolerant work space. We also derive design templates for spatial fault tolerant manipulators. When joint limits are introduced, analytic solutions become infeasible but instead a numerical design procedure can be used, as is illustrated through an example.>

A new approach for robotic visual tracking and servoing is discussed. Deformable active models are introduced as a powerful means for tracking a moving rigid object (an eye-in-hand robot arm is used). Deformable models imitate, in real-time, the dynamic behavior of elastic structures. These computer-generated models are designed to capture the silhouette of rigid objects with well defined boundaries in terms of image gradient. By means of an eye-in-hand robot arm configuration, the desired motion of the end-effector is computed with the objective of keeping the target's position and shape invariant with respect to the camera frame. Experimental results are presented for the tracking of a rigid object moving in the three-dimensional space.>

Presents a coherent approach for visual servoing tasks that consists of following the contours of unknown objects using a hand-eye robotic system. The motion of the contour relative to the camera sensor is estimated in real-time by processing the measured optical flow of a set of relevant feature points. The desired motion of the end-effector is then computed to minimize the displacements of the feature points with respect to a reference configuration in the image. Control schemes are used in order to stabilize the robot and enhance tracking performances. The authors illustrate the approach by experimental results acquired with a real-time environment, where their CMU-Direct-Drive Arm II follows the boundaries of motionless objects positioned in a plane parallel to the image plane. Three degrees of freedom of planar motions are thus servoed to perform the contour-following task.

The authors discuss how to design a manipulator from a given task. While the ultimate goal is to design all kinematic and dynamic parameters of a manipulator, they consider only kinematic parameters. Optimal design of a manipulator even for a simple task is difficult because it involves a complex task description and constraints on possible manipulators and many design variables that lead to a combinatorial explosion. For the example of designing a space shuttle tile servicing robot, they introduce a formulation of a given task and manipulator constraints.

High level assembly plans prescribe the sequence in which parts come together along with their motions. At each stage of assembly, the stability of subassemblies is an important concern. The authors address the problem of the gravitational stability of assemblies of frictionless rigid bodies. Solution methods to the problems of determining if an assembly is stable, and finding a stable orientation for a given for a given assembly are proposed. The solution methods for the two problems compute constrained motions of individual parts that decrease the gravitational potential energy, in order to determine the stability and search for stable orientations. Linear programming is used to obtain numerical solutions. The problem of finding a stable orientation is formulated as a maximin problem. The solution to this problem is the first general method for automatically determining stable orientations. The program to solve the two problems takes as input a geometric model of the assembly and information related to grounding constraints. Preliminary results are presented.

The authors present algorithms for robotic (eye-in-hand configuration) real-time visual tracking of arbitrary 3D objects traveling at unknown velocities in a 2D space (depth is given as known). Visual tracking is formulated as a problem of combining control with computer vision. A mathematical formulation of the control problem that includes information from a novel feedback vision sensor and represents everything with respect to the camera frame is presented. The sum-of-squared differences (SSD) optical flow is used to compute the vector of discrete displacements each instant of time. These displacements can be fed either directly to a PI (proportional-integral) controller or to a pole assignment controller or discrete steady-state Kalman filter. In the latter case, the Kalman filter calculates the estimated values of the system's states and the exogenous disturbances, and a discrete LQG (linear-quadratic Gaussian) controller computes the desired motion of the robotic system. The outputs of the controllers are sent to the Cartesian robotic controller. Performance results are presented.>

A method to determine constraints on translational and rotational motion of planar and 3-D objects from their contact geometry is presented. Translations are represented by spatial vectors and rotations by axes in space. For each of these, a geometric realization (M/sub a/) of the space of motion parameters is created. Subspaces in M/sub a/ that represent the range of values of motion parameters that are disallowed due to the contact are identified. The geometric realization makes it easier to visualize results, provides a good measure of the extent of restraints between objects, reduces computations by eliminating redundant constraints, and simplifies computation of new constraints. The proposed representation can be used effectively to automate the evaluation of motion constraints.>

The authors address the problem of integrating the human operator with autonomous robotic visual tracking and servoing modules. A CCD (charge coupled device) camera is mounted on the end-effector of a robot and the task is to servo around a static or moving rigid target. In manual control mode, the human operator, with the help of joystick and a monitor, commands robot motions in order to compensate for tracking errors. In shared control mode, the human operator and the autonomous visual tracking modules command motion along orthogonal sets of degrees of freedom. In autonomous control mode, the autonomous visual tracking modules are in full control of the servoing functions. Finally, in traded control mode, the control can be transferred from the autonomous visual modules to the human operator and vice versa. The authors present an experimental setup where all these different schemes have been tested. Experimental results of all modes of operation are presented and the related issues are discussed. In certain degrees of freedom the autonomous modules perform better than the human operator. On the other hand, the human operator can compensate fast for failures in tracking while the autonomous modules fail.>

The authors present results of experimental tests of a broad spectrum of basic force control strategies: proportional with feedforward, integral, filtered feedback proportional-derivative, and second-order low-pass filtering. The experiments were performed on the CMU Direct Drive Arm II so that the results would reflect the nature of the controller and not the unmodeled aspects of the system which comprises the plant. The commonality amongst the experiments was such that it was possible to compare and contrast the strategies objectively and to draw conclusions about the efficacy of each. The results support previous analysis and show the superiority of integral force control for force trajectory tracking.>

This paper presents a new approach for visual tracking and servoing in robotics. We introduce deformable active models as a powerful means for tracking a rigid object in movement within the manipulator's workspace. Deformable models imitate, in real-time, the dynamic behavior of elastic structures. These computer-generated models are designed to capture the silhouette of rigid objects with well-defined boundaries, in terms of image gradient. By means of an eye-in-hand robot arm configuration, the desired motion of the end-effector is computed with the objective of keeping the target's position and shape invariant with respect to the camera frame. Optimal estimation and control techniques (ZQG regulator) have been successfully implemented in order to deal with noisy measurements provided by our vision sensor. Experimental results are presented for the tracking of a rigid object moving in a plane parallel to the image plane (three degree-of-freedom visual servoing). The experiments performed in a real-time environment show the effectiveness and robustness of the proposed method for servoing tasks based on visnal feedback control.

In this paper, we introduce a new algorithm called Multi-Population Genetic Algorithm (MPGA) as an effi- cient optimization technique for highly nonlinear problems. Our MPGA is a parallel implementation of a GA and shows its robustness for our problem domain (Task Based Design; TBD) where we design a manipulator which is best suited for a given task. The MPGA has the same number of optimization fuac- tions as the number of task points and maintains almost con- stant complexity and does not depend on the number of task points. In addition, we develop a framework called Progressive Design, in which we design progressively based on coarse-6ne approach.

Abstract—In this paper we develop a framework for integrating real-time software modules that comprise a reconfigurable multisensor based system. Our framework is based on the proposed concept of a global database of state information through which realtime software modules exchange information. This methodology allows the development and integration of reusable software in a complex multiprocessing environment. A reconfigurable sensor-based control system consists of many software modules, each of which can be modelled using a simplified version of a port automaton. Our new state variable table mechanism can be used in both statically and dynamically reconfigurable systems, and it is completely processor independent. Individual modules may also be combined into larger modules to aid in building large systems, and to reduce bus and CPU utilization. An efficient implementation of the state variable table mechanism, which has been integrated into the Chimera II Real-Time Operating System, is also described. 1 Keywords—reconfigurable sensor-based control systems, reusable software, real-time operating systems, interprocessor communication, spin-locks, real-time software modelling. I.

This paper presents a complete overview of basic strategies that have been proposed for force control of robot manipulators. First, the model of the plant to be controlled is reviewed. Next, the strategies are divided into force-based and position-ba9ed categories, according to previously reported implementations. Each of the controller types within these categories are analyzed, and predictions of stability and efficacy are made. Then it is shown that these two categories are actually the same, and this recognition leads to the concept of a novel second order low pass filter controller. All of the predictions that result from this analysis have been experimentally confirmed. Among the contributions presented here is a comprehensive explanation of why integral gain control is the best method of force control for manipula t om.

Using active monocular vision for 3-D visual control tasks is difficult since the translational and the rotational degrees of freedom are strongly coupled. The paper addresses several issues in 3-D visual control and presents adaptive control schemes for the problem of robotic visual servoing (eye-in-hand configuration) around a static rigid target. The objective is to move the image projections of several feature points of the static rigid target to some desired image positions. The inverse perspective transformation is assumed partially unknown. The adaptive controllers compensate for the servoing errors, the partially unknown camera parameters, and the computational delays which are introduced by the time-consuming vision algorithms. The authors present a stability analysis along with a study of the conditions that the feature points must satisfy in order for the problem to be solvable. Finally, several experimental results are presented to verify the validity and the efficacy of the proposed algorithms.>

A novel formulation of the artificial potential approach to the obstacle avoidance problem for a mobile robot or a manipulator in a known environment is presented. Previous formulations of artificial potentials for obstacle avoidance have exhibited local minima in a cluttered environment. To build an artificial potential field, the authors use harmonic functions that completely eliminate local minima even for a cluttered environment. The panel method is used to represent arbitrarily shaped obstacles and to derive the potential over the whole space. Based on this potential function, an elegant control strategy for the real-time control of a robot is proposed. Simulation results are presented for a bar-shaped mobile robot and a three-degree-of-freedom planar redundant manipulator.>

The Chimera II has been designed as a local operating system, to be used in conjunction with a global operating system. It executes on one or more single board computers in a VMEbus-based system. Advanced sensor-based control systems are both statistically and dynamically reconfigurable. As a result, they require many special features, which are currently not found in commercial real-time operating systems. Several design issues for such systems are presented as well as the features the authors have developed and implemented as part of Chimera II. These features include: a real-time kernel with dynamic scheduling, global error handling, user signals, and two levels of device drivers; an enhanced collection of interprocessor communication mechanisms, including global shared memory, spin-locks, remote semaphores, priority message passing, global state variable tables, multiprocessor servo task control, and host workstation integration; and several support utilities, including a UNIX C and math libraries, a matrix library, a command interpreter library, and a configuration file library.>

A formulation of the artificial potential approach to the obstacle avoidance problem for a mobile robot or a manipulator in a known environment is presented. Previous formulations of artificial potentials, for obstacle avoidance, have exhibited local minima in a cluttered environment. To build an artificial potential field, harmonic functions that completely eliminate the local minima even for a cluttered environment were used. The panel method was used to represent arbitrarily shaped obstacles and to derive the potential function over the whole space. Based on this potential function and elegant control strategy for the real-time control of a robot is proposed.>

Algorithms for robotic real-time visual tracking of arbitrary 3-D objects traveling at unknown velocities in a 2-D space are presented. The problem of visual tracking is formulated as a problem of combining control with computer vision. A mathematical formulation that is general enough to be extended to the problem of tracking 3-D objects in 3-D space is presented. The authors propose the use of sum-of-squared differences optical flow for the computation of the vector of discrete displacements each instant of time. These displacements can be fed either directly to a PI controller, a pole assignment controller, or a discrete steady-state Kalman filter. In the latter case, the Kalman filter calculates the estimated values of the system's states and exogenous disturbances, and a discrete LQG controller computes the desired motion of the robotic system. The outputs of the controllers are sent to a Cartesian robotic controller that drives the robot.>

The authors present experimental results from a strategy for stable hard-on-hard impacts of a robot manipulator, and show how this method fits into the framework of force control. Previously proposed methods of impact control are reviewed, and their inadequacies are revealed. A fourth-order model of an arm/sensor/environment system is reviewed. It is pointed out how proportional gain force control with feedforward and impedance control are equivalent when the environment is stiff. An analysis of the model and the controllers shows the gain selection that provides the best impact response. Extensive experimental results confirm the correctness of the model, the equivalence of the controllers, and the efficacy of this impact control strategy.>

Presents strategies for reducing the impact force resulting from the collision of a kinematically redundant manipulator with an object in its environment. The Premultiplier Diagram, a tool used to derive the impact force reduction strategies presented, is introduced and discussed. Two strategies for reducing impact force are then presented. The first strategy involves adding torques to the joints of the redundant manipulator to impede motion into the object with which it collides. The second strategy involves choosing the best configuration for the impact event. Simulated results from the testing of either strategy are presented and discussed.>

The authors investigate dexterity measures of a manipulator based on its Jacobian matrix. The goal is to derive dexterity measures which can be used for both design and control of a manipulator. Generally, a dexterity measure must be independent of the scale of a manipulator for design, and must be expressed analytically so that it can be used for real-time control. Every dexterity measure must bear a physical meaning. The measure of manipulability has an analytical expression, but it depends on the scale of a manipulator. On the other hand, the condition number is independent of the scale, but cannot be expressed analytically. These two main problems (scale dependency and analytical expression) of previous dexterity measures derived from the Jacobian matrix are solved and applied in design and control of manipulators. In addition, a dexterity measure called the measure of isotropy is introduced.>

The planning problem for a mobile manipulator system that must perform a sequence of tasks defined by position, orientation, force, and moment vectors at the end-effector is considered. Each task can be performed in multiple configurations due to the redundancy introduced by mobility. The planning problem is formulated as an optimization problem in which the decision variables for mobility (base position) are separated from the manipulator joint angles in the cost function. The resulting numerical problem is nonlinear with nonconvex, unconnected feasible regions in the decision space. Simulated annealing is proposed as a general solution method for obtaining near-optimal results. The problem formulation and numerical solution by simulated annealing are illustrated for a manipulator system with three degrees of freedom mounted on a base with two degrees of freedom. The results are compared with results obtained by conventional nonlinear programming techniques customized for the particular example system.>

The planning problem is considered for a mobile manipulator system which must perform a sequence of tasks defined by position, orientation, force, and moment vectors at the end effector. Each task can be performed in multiple configurations due to the redundancy introduced by mobility. The planning problem is formulated as an optimization problem in which the decision variables for mobility (base position) are separated from the manipulator joint angles in the cost function. The resulting numerical problem is nonlinear with nonconvex, unconnected feasible regions in the decision space. Simulated annealing is proposed as a general solution method for obtaining near-optimal results. The problem formulation and numerical solution by simulated annealing are illustrated for a positioning system with five degrees of freedom. These results are compared with results obtained by conventional nonlinear programming techniques customized for the particular example system.>

A description is given of the CHIMERA II programming environment and operating system, which was developed for implementing real-time robotic systems. Sensor-based robotic systems contain both general- and special-purpose hardware, and thus the development of applications tends to be a time-consuming task. The CHIMERA II environment is designed to reduce the development time by providing a convenient software interface between the hardware and the user. CHIMERA II supports flexible hardware configurations which are based on one or more VME-backplanes. All communication across multiple processors is transparent to the user through an extensive set of interprocessor communication primitives. CHIMERA II also provides a high-performance real-time kernel which supports both deadline and highest-priority-first scheduling. The flexibility of CHIMERA II allows hierarchical models for robot control, such as NASREM, to be implemented with minimal programming time and effort.>

A potential function based on superquadrics is presented that closely models a large class of object shapes. This potential function also prevents the creation of local minima when it is added to spherically symmetric attractive wells. Two compatible forms of the superquadric potential function are introduced: one for obstacle avoidance, and another for obstacle approach. The avoidance and approach potentials are implemented in simulations. In these simulations the end effector of the manipulator experiences an attractive force from a global spherical well, while the end effector and each of the links experience repulsive forces from all of the objects. The authors have also experimentally implemented the avoidance potentials on the CMU DDARM II system. The results demonstrate successful obstacle avoidance and approach, and exhibit an improvement over existing schemes.>

CHIMERA is a real-time computing environment used in the Reconfigurable Modular Manipulator System project. CHIMERA, which is both a hardware and software environment, allows rapid development and implementation of real-time control programs. It provides a C/Unix-flavored concurrent programming environment for a Motorola 68020 multiprocessor hardware configuration connected to a Sun workstation. CHIMERA has been implemented using commercial hardware in conjunction with a sophisticated, locally developed software package, resulting in a reliable, reasonably priced, and easily duplicated system. CHIMERA is currently being ported for real-time control of the CMU Direct Drive Arm II. The authors describe the implementation and capabilities of the CHIMERA environment and illustrate how these features are used in robot control applications.>

The author addresses the problem of generating persistently exciting trajectories for parameter estimation. To achieve this, he proposes an algorithm that categorizes the dynamics parameters of a manipulator into three classes: uniquely identifiable, identifiable in linear combinations only, and unidentifiable. Since the proposed algorithm categorizes the parameters only on the basis of the kinematic configuration of the manipulator, it makes it possible to determine if a chosen trajectory (for estimation) is persistently exciting. The categorization procedure also helps in making the estimation procedure more robust. Based on the proposed algorithm, the parameters of the CMU DD (Carnegie-Mellon University Direct-Drive) Arm II were categorized.>

Experimental results on the real-time performance of model-based control algorithms are presented. The computed-torque scheme which utilizes the complete dynamics model of the manipulator was compared to the independent joint control scheme, which assumes a decoupled and linear model of the manipulator dynamics. The two manipulator control schemes have been implemented on the Carnegie-Mellon University DD (direct-drive) Arm II with a sampling period of 2 ms. The authors discuss the design of controller gains for both the computed-torque and the independent joint control schemes and establish a framework for comparing their trajectory-tracking performances. It is shown that the computed-torque scheme outperforms the independent joint control scheme as long as there is not torque saturation in the actuators. Based on the experimental results, the authors conclusively establish the importance of compensating for the nonlinear Coriolis and centrifugal forces even at low speeds of operation.>

An algorithm is proposed that automatically generates the Denavit-Hartenberg (DH) kinematic parameters of a reconfigurable manipulator. The DH kinematic parameters are then used to obtain the forward kinematic transformation of the system. The authors also address the problem of obtaining the inverse kinematics of reconfigurable manipulators. To automate the inverse kinematics and to make the procedure as general as possible, they use a numerical approach. In the case of a redundant manipulator they utilize the extra degrees of freedom to achieve singularity avoidance.>

An overview is presented to research on the analysis, synthesis, real-time implementation and performance evaluation of model-based manipulator control schemes. The schemes synthesize strategies that include a dynamical model of the manipulator in the feedback loop. Depending on the way that the dynamical model is incorporated, it is possible to create different types of control laws. For example, the computed-torque method utilizes the model is the feedback loop in order to both decouple and linearize the system. Independent joint controllers are then designed to achieve accurate trajectory tracking and to reject unknown external disturbances. The feedforward control scheme is another model-based control method and utilizes the dynamics model in the feedforward path. The idea is that the feedforward torques/forces provide gross signals and independent joint controllers provide the correcting control signals to reject disturbances that are unknown.>

In a previous paper (1987), the authors proposed a parallel computational scheme that is based on the mathematical decomposition of the equations into their primitive matrix/vector arithmetic operations. It was shown that the mathematical decomposition scheme provides an efficient mechanism to reduce the computational cycle of both the Newton-Euler (N-E) and the Lagrange-Euler (L-E) formulations. In the present paper, the N-E and L-E equations are analyzed from a hardware perspective and the results for each are compared. The analysis shows that N-E is more efficient than L-E from the computational as well as the hardware point of view.>

An obstacle-avoidance potential based on superquadrics is discussed. The superquadric formulation is a generalization of the elliptical potential function method and therefore is viable for a much larger class of object shapes. As with elliptical potentials, a modified form of the superquadric potential provides safe approach objects. The avoidance and approach potentials are implemented in simulations and the results exhibit an improvement over existing potential schemes. The simulations also use an algorithm that eliminates collisions with obstacles by calculating the repulsive forces exerted on links, based on the shortest distance to an object.>

In our previous research, we experimentally implemented and evaluated the effect of dynamics compensation in model-based control algorithms. In this paper, we evaluate the effect of changing the control sampling period on the performance of the computed-torque and independent joint control schemes. While the former utilizes the complete dynamics model of the manipulator, the latter assumes a decoupled and linear model of the manipulator dynamics. We discuss the design of controller gains for both the computed-torque and the independent joint control schemes and establish a framework for comparing their trajectory tracking performance. Our experiments show that within each scheme the trajectory tracking accuracy varies slightly with the change of the sampling rate. However, at low sampling rates the computed-torque scheme outperforms the independent joint control scheme. Based on our experimental results, we also conclusively establish the importance of high sampling rates as they result in an increased stiffness of the system.

This paper presents the experimental results of the real-time performance of model-based control algorithms. We compare the computed-torque scheme which utilizes the complete dynamics model of the manipulator with the independent joint control scheme which assumes a decoupled and linear model of the manipulator dynamics. The two manipulator control schemes have been implemented on the CMU DD Arm II with a sampling period of 2 ms. Our initial investigation shows that the computed-torque scheme outperforms the independent joint control scheme as long as there is no torque saturation in the actuators.