We present the ANBOT, an intelligent robotic coworker for physical human-robot collaboration. The ANBOT system assists workers performing industrial abrasive blasting, shielding them from the large forces experienced during this physically demanding task. The co-operative robotic system combines the strength and endurance of robots with the decision making of skilled workers. The inherent challenges in human-robot collaboration, combined with the difficult blasting environment required novel design decisions to be made and new solutions to be developed. These include an approach for handling kinematic singularities in a manner suitable for human-robot co-operation, estimating worker pose under poor visibility conditions, and an intuitive control scheme that adapts the robotic assistance based on the estimated strength of the worker. In this work we summarise the ANBOT system and present findings from preliminary site trials. The trials included several real industrial blasting tasks under the control of a skilled abrasive blasting worker who had no experience working alongside a robot. Results demonstrate the suitability of the ANBOT for practical industrial applications.

This paper presents a new configuration for single axis tactile sensor arrays molded in rubber to enable tri-axial force measurement. The configuration requires the sensing axis of each sensor in the array to be rotated out of alignment with respect to external forces. This angled sensor array measures shear forces along axes in a way that is different to a planar sensor array. Three sensors using the angled configuration (22.5°, 45° and 67.5°) and a fourth sensor using the planar configuration (0°) have been fabricated for experimental comparison. Artificial neural networks were trained to interpret the external force applied along each axis (X, Y and Z) from raw pressure sensor values. The results show that the angled sensor configuration is capable of measuring tri-axial external forces with a root mean squared error of 1.79N, less error in comparison to the equivalent sensor utilizing the planar configuration (4.52N). The sensors are then implemented to control a robotic arm. Preliminary findings show angled sensor arrays to be a viable alternative to planar sensor arrays for shear force measurement; this has wide applications in physical Human Robot Interaction (pHRI).

Robotic systems such as exoskeletons can be effectively used in the reduction of fatigue and musculoskeletal disorders (MSD) associated with physical tasks, but robots which work in physical contact with humans pose problems with user safety. A novel approach to developing intrinsically safe robots is to use passive actuators which have the advantage of being safer, ensuring stability, high force/weight ratios and lower power consumption. It is however not clear how effective an exoskeleton utilizing passive actuators would be in reducing fatigue and the risk of MSD. This paper analyzes the benefit of using such a system with results from dynamic simulations and an experiment using a specially designed mechanism used for evaluation. Results indicate that fatigue and effort could be reduced if robot impedance is minimized. Experiments also highlighted issues of implementing such a system into practice.

Dynamic maneuvers have been successfully implemented on many prismatic legged robots. Systems with articulated legs of significant relative mass pose more of a challenge in part due to the physics of thrusting with rotating limbs, which results in undesired non-vertical impulses, and in part due to the control problem of synchronizing ankle and knee joints. Presented here is an experimental articulated leg system that simplifies the control of an articulated monopod through the use of a joint-coupling tendon. The ankle is coupled to the thigh with an inelastic tendon, which causes automatic horizontal impulse compensation on liftoff for varying knee thrusts. Using a tendon-coupled ankle, stable sustained hopping is achieved for a fixed-torso monopod with a very simple control strategy, and with minimal hip actuator effort, while hopping in place is achieved for short time periods with a bipedal robot with freely pitching torso.

Simulations of objects whose interaction change due to intermittent contact, collision, sticking or slipping traditionally employ mathematical models of the system in each topology. Usually these simulations rely on a tedious process of deriving separate dynamical equations for each topology and sophisticated switching logic to select the appropriate model. We introduce a unified method which requires deriving only a single set of equations and simple switching.

We are in the process of building a biped robot capable of highly dynamic maneuvers. Implementing such maneuvers in real-time entails heuristic controllers founded on a grasp of the dynamics. We present here an analysis of the mechanics of biped running and a control strategy for stable running. By expounding on the direction of the ground-reaction forces and utilizing an impulse representation of the contact phase, we present a tractable model of the mechanics. This model motivates the control strategy and provides a basis for extending the strategy to more general systems. The control strategy consists of a simple set of three rules, where the key rule considers the leg-length upon liftoff. This rule offers a simple control for both steady-state and accelerated running. We present the equilibrium running index as a characterization of biped running. It relates the parameters affecting running and has broad applicability to biological and robotic systems. The control strategy was verified in simulation and the results are presented.

Trotting and galloping allow a quadruped to rapidly traverse rough terrain. Modeling this motion, which is only dynamically stable, is of importance for legged robot operation and for quadrupedal animal motion estimation. Derived from an eight-step galloping cycle, this study presents a kinetic hybrid model in which the states vary based on the principal forces present. As compared to foot contact, or kinematic, hybrid quadruped models this reduces the maximum number of possible states from 120 to 6 and provides an alternative to foot contact monitoring. This approach was tested on a trotting quadruped robot equipped with an inertial sensor aided by video. This was processed using an EKF estimator framework to give attitude estimates at rates of up to 250 Hz with 5deg error.

An important tool in clarifying various theories governing the dynamics of rapid quadrupedal gaits, such as the trot and gallop, is the measurement of body attitude. Such measurements are complicated in open field environments because of the large ranges and high data rates needed due to the speeds (7 m/s) and rapid shifts in dynamics present. To address this a lightweight inertial sensing harness is introduced with sensing design based on the KOLT robot. Its mass center is collocated with the subject so as to reduce dynamic bias. This work combines dynamic gait system identification and motion estimation and is demonstrated on a Labrador retriever (Canis lupus familiaris) through measurements of the gallop over long spans (20 m) and at data rates comparable with gait laboratories (400 Hz). The results are consistent with laboratory measurements, but seem to suggest a roll and yaw cross-coupling during gallop.

Dynamic legged locomotion entails navigating terrain at high speed. The impact shocks from rapid footfalls, pivotal for such mobility, introduce large impulses that saturate motion measurement. A biomimetic approach is presented in which visual information, in the form of optical flow, complements information from inertial sensors. The motion is then determined using a two-phase Hybrid Extended Kalman Filter. Experimentation in determining attitudes on a robotic leg platform shows a reduction in drift over inertial approaches and in delay over visual approaches. In tests with 6g impulses, pose was recovered within 5 deg rms with angular rate errors limited to 10 deg/sec at frequencies up to 250 Hz.

Legged robots offer exceptional mobility in uncharted terrains. Their dynamic nature yields unrivaled mobility, but serves to destabilize the motion estimation process that underlies legged operations. In particular, the discontinuous foot fall patterns and flight phases result in severe impulses, which, in turn, result in excessive accumulation of drift by inertial sensors. Ground range measurements, amongst several others, are robust to this drift yet are limited in application due to their low-bandwidth and sensitivity to ground conditions. In considering the attitude estimation problem for this dynamic legged locomotion, this paper develops a pose calculation method based on ground range measurements. This is used in conjunction with a hybrid Extended Kalman Filter that takes advantage of the ballistic nature of the flight phases. Results indicate that this combination provides rapid, robust estimates of attitude necessary for extended dynamic legged operations. In single leg experiments, which were conducted using low-cost sensing hardware, this method had an RMS error of < 1 °, half that of a non-hybrid EKF approach.

Localization and measurement of displacement are critical aspects to the operation and control of mobile robots. The motion of desktop robots, a class of mobile robots designed for use on a table, can be considered a special case of the general localization problem since the vehicle is primarily operated over a flat surface. Using the assumption that the motion of a desktop robot is essentially planar, this paper presents a novel method that senses the motion of two points to obtain both the position and orientation of the robot in two-dimensional space. The paper details how this method can be implemented using low-cost, off-the-shelf sensor hardware components and demonstrates its application in the Desktop-Bot, a compact desktop robot. Experimental testing validated features of this planar localization such as: estimation with minimal mean error (or drift), no external sensing hardware apparatus (i.e., on-board sensing), fast-update rates (/spl sim/50 Hz) and robustness to external occlusion.

Intelligent controllers are being used with increasing effectiveness on complex systems. This work verifies the effectiveness of fuzzy control, an intelligent method, on a single, articulated-leg that was designed to be used on a high-speed galloping quadruped. Intelligent methods are compared to other control methods in simulation and on the OSU DASH (Dynamic Articulated Structure for High-performance) leg. It is shown that the intelligent controllers outperform non-learning methods. Using fuzzy control, the OSU DASH leg performs stable hopping on a treadmill moving at 2.0 m/s.

In this paper we describe a software system built to coordinate an autonomous vehicle with variable configuration ability operating in rough terrain conditions. The paper describes the system architecture, with an emphasis on the action planning function. This is intended to work with a proprioceptive algorithm that continuously coordinates wheel torques and suspension forces and positions to achieve optimal terrain crossing performance.

Media-rich learning experiences are widely available on the web.Medical schools and other health care organizations maintain extensive websites for learning, reference, and assessment.Continuing medical education on the Internet is one of the most commonly available and widely used services for health care practitioners.These media-based learning environments, though widely available, represent a passive medium of information retrieval by clicking a mouse.We have been investigating the next generation of Internet-based learning technologies: simulation-oriented learning environments that support interaction, collaboration, and active learning.The topics of learning are anatomy and basic surgical manipulations.To support these topics, we have developed three-dimensional visualizations of anatomy and have enriched these anatomical models with biomechanical tissue properties of elasticity and viscosity such that the learner can use appropriate force-feedback tools to feel tissues at the same time that they visualize them.Access to and interaction with these simulation environments have required that we push the envelope of current Internet capabilities, and that we investigate the role of network bandwidth and latency in the performance of these applications.A preliminary version of this work was presented at MedInfo, 2001. 1 BackgroundThe learning of anatomy and surgery involves a visuo-haptic-audio experience.While books, lectures, and multimedia are important routes to this learning, the acts of touching, feeling and cutting are believed to be essential in the training of new physicians and surgeons.Simulated environments that deliver this experience are expected to form the next generation of technology-enhanced learning environments.When this research was begun in 1998, simulators were in development for knee arthroscopy, laparoscopy, endoscopy, epidural needle insertion, and sinus endoscopy.2-10 These were three-dimensional graphical simulations of anatomy, with haptic feedback, to support planning a surgical or clinical procedure.Some included the ability to cut and move tissue.11 Others

Advanced collaborative display systems allow users to view a computing desktop environment in a platform- and location-independent fashion. For real-time considerations, these systems become computationally very challenging, especially when video streaming is included. The addition of stereoscopic video streaming is desirable in virtual environments (VE) created for the teaching of anatomy and surgery with real-time collaborative audio and video interactions at many locations. However, this stresses the real-time requirements to the point at which realistic video is difficult to assure. For such a system to work, it is imperative that timing data be collected, analyzed and understood. We describe an experimental system designed primarily for the collection of timing data that is required for robust collaborative medical training applications. The networked stereoscopic system uses a server-swappable multicast network protocol to stream real-time manipulations of 3D virtual body structures at the server site to all clients participating in the multicast session. The three visual modes have dynamic tuning parameters for adjusting the parallax and the focal point for the rendered scene, allowing users to define individual stereoscopic comfort zones. Optimizing the graphics module is critical in achieving the necessary rendering rates. Different techniques of utilizing various memory resources increased the number of polygons rendered per second by over seven million. Depending on the type of memory used, the number of polygons rendered per second varies from 2.25 to 9.12 million.

In this article we discuss some of the important similarity rules, or scaling laws that apply to robotic mechanisms. The important question is: "What information can I derive about a smaller or larger system from a similarly configured, operational system?" The scaling laws elucidated here are useful for the initial configuration of new designs.

In this paper, an efficient and systematic formulation of the force distribution equations for general tree-structured robotic mechanisms is presented. The applicable platforms include not only systems with star topologies, such as walking machines that have multiple legs with a single body but also general tree-structured mechanisms, such as variably configured wheeled vehicles having multiple modules. The force balance equations that govern the relationship between the contact forces and the resultant inertial forces/moments of the vehicle will be derived through a recursive and computationally efficient algorithm. Also, the joint torque constraints that specify the joint actuator limits, and contact friction constraints that may be used to avoid slippage and maintain contact, are efficiently incorporated in the formulation. Based on this formulation, several standard optimization techniques, such as linear programming or quadratic programming, can be applied to obtain the solution. An algorithm summarizing the results developed, and suitable for computer implementation, is included. The algorithm has been applied to an n-module actively articulated wheeled vehicle, and the computational cost evaluated. The efficiency of the algorithm is demonstrated with results showing real-time execution on a Pentium PC.

An efficient formulation of the force distribution equations for actively-coordinated vehicles is presented. The applicable platforms include not only systems with star topologies, such as walking machines that have multiple legs with a single body, but also general tree-structured mechanisms, such as variably-configured wheeled vehicles having multiple modules. Based on this formulation, several standard optimization techniques, such as linear programming or quadratic programming, can be applied to obtain the solution. The efficiency of the formulation is demonstrated with results showing real-time execution on a Pentium PC.

Combined grasping and manipulation by multifingered hands was studied. A method of force allocation by optimizing the friction angles at finger contacts was combined with the computed torque method to find the torques to be commanded at finger joints for multifingered systems. In this way, the chance of slip can be minimized when the object is grasped or manipulated. A rolling motion between the grasped object and the fingertips was studied in simulation. A history-based method was proposed to improve the smoothness of the input torque commands, and force feedback was used to compensate for the uncertainty of the system. Any finger which moved beyond the boundaries of its working volume was also moved to a new position for regrasping. The proposed method can be efficiently used to find the necessary actuator torques, and is applicable for real-time application. Three dimensional simulation results are given in this paper.>

The dependency of payload on vehicle operating speed, which is a problem peculiar to legged locomotion, is examined via a case study of a hexapod using wave gaits. The treatment presented is confined to systems using quasi-statically stable gaits. A set of closed-form analytic expressions describing the payload-speed relationship has been obtained. The solution is based on the assumption of equal leg compliances. The utility of such relationships lies in determining the allowable payload given the vehicle speed and weight or, conversely, in preplanning the payload to be carried and determining the maximum speed at which the vehicle can operate safely.>

The Stewart platform is a fully parallel, six-degree-of-freedom manipulator mechanism. Although its inverse kinematics have been extensively studied, no solutions to the direct position kinematics problem have been previously presented in the literature. A solution of the direct kinematics problem of the case in which the six limbs form three concurrent pairs at either the base or the hand member is presented. Even though it is not the most general possible configuration, this case does include many arrangements that have been used in practical robot mechanisms.>

The Stewart platform is a fully parallel, six-degree-of-freedom manipulator mechanism. A solution of the direct kinematics problem for the Stewart platform is presented for the case in which the six limbs form three concurrent pairs at either the base or the hand member. The solution is shown to be reducible to a 24th-order polynomial equation in tan( phi /sub r//2). It is also shown that, for a given set of link lengths, the Stewart platform can be assembled in 24 different configurations.>

A method is presented which addresses the problem of determinating the appropriate distribution of forces among the fingers of a multifingered gripper grasping an object. The finger-object interactions are modeled as point contacts. The system is statically indeterminate, and an optimal solution for this problem is desired for force control. A fast and efficient suboptimal method for computing the grasping forces is presented. In addition to determining the equilibrating forces (the forces required to maintain equilibrium), it is essential to superpose interaction forces (the forces which squeeze the object) to ensure that the fingers do not slip. The decomposition of the equilibrating forces into forces perpendicular and parallel to the load wrench provides a useful simplification. If it is assumed that the normals to the object at the point of contract pass through the centroid of the contact points, the computations required to find the interaction forces are considerably simplified. Some simple grasps are used to evaluate the proposed algorithms.>

The problem of force distribution in systems involving multiple frictional contacts between actively coordinated mechanisms and passive objects is examined. The special case in which the contact interaction can be modeled by three components of forces (zero moments) is particularly interesting. The Moore-Penrose generalized inverse solution for such a model (point contact) is shown to yield a solution vector such that the difference between the forces at any two contact points projected along the line joining the two points vanishes. Such a system of forces is described by a helicoidal vector field which is geometrically similar to the velocity field in a rigid body twisting about an instantaneous screw axis. A method to determine this force system is presented. The possibility of superposing another force field which constitutes the null system is investigated.>

The problem of force distribution in systems involving multiple frictional contacts between actively coordinated mechanisms and passive objects is examined. The special case in which the contact interaction can be modeled by three components of forces (zero moments) is particularly interesting. The Moore-Penrose generalized inverse solution for such a model (point contact) is shown to yield a solution vector such that the difference between the forces at any two contact points projected along the line joining the two points vanishes. Such a system of contact forces is described by a helicoidal vector field which is geometrically similar to the velocity field in a rigid body twisting about an instantaneous screw axis. A method to determine this force system is presented. The possibility of superposing another force field which constitutes the null system is also investigated.>

For legged locomotion the ability to preplan allowable payload given specified operating conditions is important, both in design and high level task planning applications. In this paper, the dependency of payload on vehicle operating speed, a problem peculiar to legged locomotion, is examined via a case study of a hexapod using wave gaits. As a result, a set of closed form analytic expressions describing the payload-speed relationship has been obtained. The solution is based on the assumption of equal leg compliances. The utility of such relationships lies in determining the allowable payload given the vehicle speed and weight or, conversely, in preplanning the payload to be carried and determining the maximum speed at which the vehicle can operate safely.

The work described in this paper addresses the problem of determination of the appropriate distribution of forces between the fingers of a multifingered gripper grasping an object. The system is statically indeterminate and an optimal solution for this problem is desired for force control. A fast and efficient sub-optimal method for computing the grasping forces is presented. This method is based on the superposition of finger-interaction forces on equilibrating forces. An interaction force is defined as the component of the vector difference of the finger contact forces at any two fingers along the line joining the two contact points. They are computed based on the assumption that the normals at the point of contact pass through the centroid of the contact points and are therefore independent of the actual geometry of the object. The contact interaction is modelled as a point contact. The problems associated with making the algorithm independent of the object geometry are explored.

A proximity sensor using a mechanical contact principle is under development for robotic applications. Good discrimination between contact loads and inertial windage loads is essential. Rugged design and easy replacement of sensing elements is also necessary. The sensor is under consideration for use on the feet of the Adaptive Suspension Vehicle.

Six degrees of freedom is only the minimum requirement for a manipulator to have full spatial positioning capability. However, effective use of more than six degrees of freedom is not a simple problem. Those systems with redundant degrees of freedom which have been successfully operated are based on end-point sensing. The present paper deals with ways of making effective use of additional degrees of freedom in conventional, open-loop, industrial robot type operation.

The control of a legged locomotion system over uneven terrain requires a new approach to coordination. The system has a high degree of static indeterminacy and, because of the stiffness of the structural and actuation systems, has stability problems if position-velocity control is used. The problem has many similarities to that of grasping and manipulating an object in a multi-fingered hand. The stability problem is removed by controlling actuator force directly, but the necessary computations can be prohibitive. A solution is presented which is moderate in its computational requirements. It is based on an assumption of zero foot interaction forces in the horizontal plane.

Broadly speaking, the regional structure of a manipulator, which consists of the inboard three joints and the members associated with them, determines the workspace shape and volume. The orientation structure, which, for a six degree of freedom manipulator, consists of the three furthest outboard joints and members, determines the geometric dexterity or orientation potential of the manipulator. It is possible, using straightforward geometric arguments, to determine the optimal dimensions of the regional structure for a given total length. By the use of the spherical counterpart of Grashof's theorem formulated by Freudenstein, it is possible also to show that there is an optimum geometry of the orientation structure.

Mobile platforms available for use in robotic systems vary enormously in their mobility and in their control characteristics. In particular, the characteristics which are desirable in an autonomous vehicle for unstructured terrain are quite different from those which are optimum in a vehicle for partially structured terrain, such as road following, or for a mobile platform for use in a highly structured factory environment. The controlled degrees of freedom of the platform impact the manner in which it can be used. Platforms with articulated steering are severely limited in situations, such as mobile industrial robot service, in which lateral motion may be desirable.