The paper develops a control scheme for task-space trajectory tracking of robot manipulators in the face of unknown kinematic and dynamic parameters of both the manipulator and picked object/tool, and in the presence of time varying disturbances/noise and certain unmodelled dynamics. The proposed scheme consists of an outer task-space control loop and an inner joint space control loop. The two control loops in combination ensure that the errors tend to zero asymptotically and thus achieve precise task-space tracking. Other features of the proposed method are simple control structure that do not require decomposition of the dynamics into the regressor matrix-parameter vector. Simulation results are provided for a full 6-DOF manipulator performing complex tasks while carrying a tool with unknown kinematics and dynamics parameters.

The paper develops a unified kinematics modelling, optimization and control for hybrid robots. These robots combine two or more modes of operations, such as a combination of walking and rolling, or rolling and manipulation. The equations of motion are derived in compact forms that embed an optimization criterion. These equations are used to obtain various useful forms of the robot kinematics. Using the developed modelling, actuation and control equations are derived that ensure the robot to track a desired path closely while maintaining balanced operations and tip-over avoidance. Various simulation results are provided for a hybrid rolling-walking robot traversing uneven terrain, which demonstrate the capabilities and effectiveness of the developed methodologies.

The paper proposes a real-time robotic person following using a combination of shape detection and color histogram matching with the adaptation of the search area. First the person's head is detected in the image by applying Canny's edge detection and a modified Hough transform. Next the area under the head in the image is searched for the person's shirt color characteristics. The search area in the image for finding the person is adaptively expanded or contracted to ensure fast computation and low false detection. The proposed image processing techniques together with the adaptation achieves a reliable real-time system for fast person following. The system has been implemented on a Segway robotic platform and results are provided.

Serial kinematic chains are widely used in robotics and computer animation among other fields. Many manipulators do not have closed-form solutions to the inverse kinematics problem, which is of great importance for many applications. In this paper we introduce a fast and accurate procedure which yields all joint angle solutions for a given manipulator or limb posture (position and orientation) and certain swivel angle. By means of a spatial decomposition method, the procedure involves finding accurate models which represent the behavior of the robot or limb in a given workspace region. We propose Support Vector Machines, a very popular machine learning method, as the method that models such behaviors. The performance of the method is tested on the Robotic Research Arm K-1207. The results confirm that the method finds accurate solutions and can be used on real world applications with real-time requirements.

This paper describes the development of an robotic helicopter for monitoring pollution and habitat in natural reserves. The particular environment under investigation is Tijuana River Estuary (TRE) south of the city of San Diego on the US-Mexico border. The ecology of TRE is briefly reviewed and the motivation for using an robotic helicopter is discussed. Next the application of aerial imaging to natural reserves is mentioned. An overview of the helicopter and its autonomy is given. The autonomous flight and capturing of images tagged with GPS and attitude are described. A method is presented to correct (rectify) aerial images that are distorted due to helicopter rolling and pitching leading to non-vertical photography. Flight test results are also provided.

A new paradigm is introduced for solving the forward kinematics of general Stewart platforms in real-time. It consists of an off-line preprocessing phase and an online realtime evaluation phase. In the preprocessing phase, the platform leg (link) space is decomposed into cells, and a large set of data is generated for the platform position/orientation (pose), and their corresponding link lengths are computed using the known inverse kinematics. Due to the existence of multiple solutions (poses) for a particular link vector, a data classification technique is employed to identify various solutions. The classified data are used to find the parameters of a simple model that represents the forward kinematics within a cell. These parameters are stored in a lookup table. During the online phase, given the link lengths, the appropriate cell is identified, the model parameters are retrieved from the lookup table and the poses are computed. The proposed method is tested on a Stewart platform, and the accuracy and online times are presented to show the effectiveness of the proposed method for real-time applications.

The paper proposes a systematic, data-driven method for kinematics modeling of high mobility wheeled rovers traversing uneven terrain. The method is based on the propagation of position and orientation velocities starting from the rover reference frame and going through various joints and linkages to the wheels. Equations of the motion are set up in a compact form, which only require the D-H parameters of the rover joints and links. To illustrate the proposed kinematics modeling, the method is applied to a rover similar to the NASA's Sample Return Rover.

The paper proposes a novel method for extremely fast inverse kinematics computation suitable for fast-moving manipulators and their path planning and for the animation of anthropomorphic limbs. In a preprocessing phase, the workspace of the robot is decomposed into small cells, and data sets for joint angle vectors (configurations) and hand positions/orientations (postures) are generated randomly in each cell using the forward kinematics. Due to the existence of multiple solutions for a desired posture, data classification is utilized to identify various solutions. The generated and classified data are used to determine the parameters of a simple linear or quadratic model that closely approximates the inverse kinematics within a cell. These parameters are stored in a lookup file. During the online phase, given the desired posture, the index of the appropriate cell is found, the model parameters are retrieved, and the joint angle vectors are computed. The advantages of the proposed method over the existing approaches are discussed. Data resulting from many trial runs are compiled for a manipulator and an anthropomorphic arm to demonstrate the performance of the proposed method.

This paper describes a general approach to kinematics modeling of all-terrain rovers traversing uneven terrain. The model is derived for full six degree of freedom motion including rotations in pitch, roll, and yaw as well as translation in x, y, and z. Differential kinematics is derived for individual wheel motions in contact with the terrain and then combined to form the composite equation for rover motion. Three useful types of kinematics (i.e., navigation, actuation and slip) are identified, and the equations and application of each are discussed. For illustration, the method is specialized to Rocky 7, a highly articulated prototype Mars rover. Simulation results are described for Rocky 7 traversing bumpy terrain.

This paper describes a general approach to the kinematics modeling and analyses of articulated rovers traversing uneven terrain. The model is derived for full 6DOF (6-degree-of-freedom) motion, enabling movements in the x,y, and z directions, as well as pitch, roll, and yaw rotations. Differential kinematics is derived for the individual wheel motions in contact with the terrain. The resulting equations of the individual wheel motions are then combined to form the composite equation for the rover motion. Three types of kinematics, i.e., navigation, actuation, and slip kinematics are identified, and the equations and application of each are discussed. The derivations are specialized to Rocky 7, a highly articulated prototype Mars rover, to illustrate the developed methods. Simulation results are provided for the motion of the Rocky 7 over several terrains, and various motion profiles are provided to explain the behavior of the rover.

Path planning is achieved by a special decomposition of the robot manipulator, an offline preprocessing stage, and a three phase online path planning scheme. The decomposition consists of separating the robot into several chains where a chain is a combination of several consecutive links and joints. Preprocessing is performed by defining a set of postures for each chain and setting up a collision table which re-integrates the chains into the full robot and stores the collision states of various discretized robot configurations with the obstacles. Path planning using a local search is performed independently in joint subspaces associated with robot chains. The paths found for the chains are synthesized to obtain a collision-free path for the robot. This decomposition reduces the exponential growth of computation with robot degrees of freedom (DOF) to that of the much lower chain DOF. As a result, it is possible to achieve short planning times for practical robots operating in three-dimensional work spaces. Analysis of computation time and space of the proposed method are presented. Results supporting the analysis are provided for a large number of path-planning trials with two practical robots operating in relatively cluttered environments.

The paper describes a method for kinematic modeling of the Rocky 7 Mars rover. The model is derived for full six degree of freedom motions enabling movements in x,y,z directions, as well as pitch, roll and yaw rotations. Forward kinematic equations are derived using wheel Jacobian matrices. The individual matrices are then combined to form the composite equation for the rover. This equation is solved to obtain the rover position and heading in terms of measured wheel velocities and certain rocker joint angles. The inverse kinematics is derived relating the individual wheel velocities and steering angles to the desired forward velocity and turning rate of the rover. Due to the special kinematic configuration of Rocky 7, a geometric approach is developed to determine the steering angles.

A simple controller is proposed for independent joint control of a manipulator to enable accurate trajectory tracking without requiring specific knowledge about the robot dynamics. Each joint controller consists of three nonlinear integral terms in the tracking error. Global asymptotic stability of tracking errors is established using the Lyapunov stability theory. The control system is shown to be robust to torque disturbances affecting the system and to a class of unmodeled dynamics. To further improve the robustness of the closed-loop system in situations where measurement noise is present, a simple modification to the controller is proposed. It is shown that exponential convergence of the error to arbitrarily small values is guaranteed with this modification. A systematic procedure for the selection of controller parameters is provided, and simulations results using the complete dynamic model of a six degree of freedom industrial robot are presented to demonstrate the excellent tracking performance of the proposed controller.

Path planning is achieved by a special decomposition of the robot manipulator, an off-line preprocessing stage and a two phase online path planning scheme. The decomposition consists of separating the robot into several chains where a chain consists of a number of consecutive links and joints. Preprocessing is performed by defining a set of postures for each chain and setting up a collision table which re-integrates the chains into full robot and specifies the collision of various defined robot postures with obstacles. In the first online phase, local planning is performed independently and simultaneously in joint subspaces associated with robot chains. In the second phase, the paths found for the chains are synthesized to obtain a collision-free path for the robot. This decomposition reduces dimensionality of the search problem substantially, and as a result it is possible to achieve short planning times for robots with many degrees of freedom. Analysis of computation time and success rate of the proposed method are presented.

A fast path planner is implemented on a SUN Sparc workstation that controls a Puma 560 manipulator. Path planning is achieved by decomposing the robot configurations into body formations and arm postures, and a two-phase path planning procedure. In the off-line planning phase, the defined formations and postures are generated, and a two dimensional collision table is set up, while in the online planning phase, a graph search is carried out to find a sequence of adjacent collision free formations and postures. The software for implementation of the path planner consists of several main modules for configuration generation, link volume modeling, obstacle data base, collision table, graphic simulator, robot interface and path finder. Extensive tests involving several thousand trial have been carried out in an environment cluttered with complex shaped obstacles such chairs, computer terminals and pipelines. The planning time and planning success rate are determined experimentally for different resolutions. The path planner finds and executes a collision free path in nearly all cases where such a path exists. The path planner is fast, and the online planning time is a couple of seconds on the average for a modified Puma 560 working in an environment containing some twenty complex shaped obstacles.

A simple decentralized controller is proposed to enable accurate trajectory tracking without requiring specific knowledge about the robot dynamics. Asymptotic stability of tracking errors is established using the Lyapunov stability theory. The control system is shown to be robust to torque disturbances affecting the system and to a class of unmodeled dynamics. The structure of the controller and the performance of the closed-loop system are analyzed. Simulations results using the complete dynamic model of a six degree of freedom industrial robot are presented to demonstrate the excellent tracking performance of the proposed controller.

The paper proposes an adaptive fuzzy control scheme for explicit force control of a robot manipulator in contact with an environment whose parameters are unknown and vary considerably. The scheme consists of three main components, a reference force model describing the desired behaviour of the force control system, a fuzzy force controller that determines the adjustment to the position control loop, and a fuzzy learning and adaptation mechanism that modifies the fuzzy force controller according to the difference between the actual and desired force responses. The modification is performed by shifting and contracting/expanding the membership functions of the fuzzy sets associated with the consequent rules of the fuzzy force controller. It is demonstrated through simulations of a manipulator that the scheme is capable of force tracking despite wide parameter variations, such as when the environment stiffness changes by several orders of magnitude.

Fast path planning is achieved by a special representation of the robot and a two-phase path planning procedure. The representation consists of defining sets of body formations and arm postures whose number depends on complexity of the environment and the required resolution of path planning. In the off-line planning phase, the defined formations and postures are generated, and a 2D collision table is set up which specifies collision with obstacles of various defined body formations and arm postures. In the online planning phase, a graph search is carried out to find a sequence of adjacent formations and postures. Because the time consuming collision checking is done off-line, the online path finding is extremely fast. Path planning is essentially performed in the work space thus avoiding the costly mapping of obstacles into the C-space. The path planner has been implemented on a SUN workstation that controls a Puma 560 manipulator.

In Su and Leung (1993), a sliding mode control method is proposed for joint trajectory tracking of robot manipulators. The method is based on the decomposition of the manipulator dynamics into the product of a constant parameter vector and a regressor matrix, originally developed by Slotine and Li (1988). This note shows that the method proposed by Su and Leung does not offer any advantage over the method of Slotine and Li.>

Fast path planning is achieved by a special representation of the manipulator and the workspace. A method is devised to work directly with ann postures instead of dealing with individual joint positions. These two aspects, namely representation and posture control, greatly speed up the path finding procedure and make it possible to perform near real-time planning in a reason- ably complex environment. The path planner uses bit maps of the arm and obstacles, and a novel search algo- rithm determines the sequence of postures the arm must take to effect a collision free path. The path planner has been implemented and tested on a PUMA manipulator in a moderately cluttered environment.

A general approach to the synthesis of model reference adaptive controllers for manipulators by means of hyperstability theory is presented. The manipulator adaptive control problem is first formulated in a form suitable for the application of hyperstability theory. The control scheme consists of feedback and cascade controllers, and an auxiliary signal. A generalized adaptation algorithm is developed which offers considerable flexibility in the choice of controller structure and its adjustable parameters.>

A decentralized model reference adaptive scheme is developed for digital control of robot manipulators. The adaptation laws are derived using hyperstability theory, which guarantees asymptotic trajectory tracking despite gross robot parameter variations. The control scheme has a decentralized structure in the sense that each local controller receives only its joint angle measurement to produce its joint torque. The independent joint controllers have simple structures and can be programmed using a very simple and computationally fast algorithm. As a result, the scheme is suitable for real-time motion control.>

The authors put forward a simple scheme for multivariable control of robot manipulators. The scheme is composed of an inner-loop stabilizing controller and an outerloop tracking controller. The inner loop utilizes a multivariable PD (proportional differential) controller to stabilize the robot by placing the poles at some desired location. The outer loop uses a multivariable PID (proportional integral differential) controller to achieve input-output decoupling and trajectory tracking. The gains of the PD and PID controllers are related directly to the linearized robot model by simple closed-form expressions. The controller gains are updated online to cope with variations in the robot model for gross motions and payload change. Alternatively, the use of high-gain controllers for gross motion and payload change are discussed.>