In this paper, experimental studies of a decentralized neural network control scheme of the reference compensation technique applied to control a 2-degrees-of-freedom (2-DOF) inverted pendulum on an x - y plane are presented. Each axis is controlled by two separate neural network controllers to have a decoupled control structure. Neural network controllers are applied not only to balance the angle of pendulum, but also to control the position tracking of the cart. The decoupled control structure can compensate for uncertainties and cancel coupling effects. Especially, a circular trajectory tracking task is tested for position tracking control of the cart while maintaining the angle of the pendulum. Experimental result shows that position control of the inverted pendulum and cart is successful.

In this paper, a novel collision avoidance algorithm of a mobile robot is proposed. Virtual force field between a robot and an obstacle is formed and regulated to maintain a desired distance. To avoid obstacles on the path of a mobile robot, intelligent hybrid force control is used to regulate force field of an accurate distance between a robot and an obstacle. Since uncertainties from robot dynamics and obstacles degrade the performance of a collision avoidance task, neural network is used to compensate for uncertainties. Simulation studies of avoiding obstacles by maintaining a desired distance on the path of a wheeled mobile robot are conducted to confirm the proposed algorithm.

In this paper, a collision avoidance algorithm using an impedance force control method for a mobile robot is proposed. A distance between a mobile robot and an obstacle is considered as a virtual force to be regulated by a well known impedance force control algorithm. The proposed collision avoidance algorithm is expected to react dynamically well when the obstacle is moving around. Simulation studies are conducted based on the dynamic model of a mobile robot. Simulation results confirm that the performance of the proposed force algorithm is better in maintaining the virtual distance to avoid moving obstacles compared with results of using a position control method.

In this paper, lateral force control of a mobile robot in contact with an object or environment is presented. Considering slip condition of wheels, the front wheel steered bicycle model of the mobile robot is derived. The explicit force control algorithm is applied to regulate contact force generated from the object which minimizes a force error between a desired force and an actual force. Performances of force tracking control for different stiffness of environment have been tested. Simulation results show that the proposed control algorithm works quite well to maintain desired contact force on the environment.

This paper presents the Simulation, Tactical Operations and Mission Planning (STOMP) software architecture and framework for simulating, controlling and communicating with unmanned air vehicles (UAVs) servicing large distributed sensor networks. STOMP provides hardware-in-the-loop capability enabling real UAVs and sensors to feedback state information, route data and receive command and control requests while interacting with other real or virtual objects thereby enhancing support for simulation of dynamic and complex events.

Describes the implementation of an Internet-based personal robot with three control modes, and experimental results on the remote control between KAIST, Korea and UC Davis, USA. The idea is to control a personal robot at the remote site (KAIST) by using a simulator provided at the local site (UC Davis). However, if the information of the current absolute position of the robot at its remote site cannot be estimated, the simulator may be useless. The absolute position of the robot can be determined by comparing a reference map with sensor information from sonar sensors and an electronic compass. A user can use three control modes-direct control mode, supervisory control mode and job scheduling mode, and monitor the current status of the robot using a graphic user interface of the simulator implemented with Java.

In this paper, the neural network force control is presented. Under the framework of impedance control, neural network is used to compensate for all the uncertainties from robot dynamics and unknown environment. A modified simple impedance function is realized after the convergence of the neural network. Learning algorithms for the neural network to minimize the force error directly are designed. As a test-bed, the large X-Y table robot was implemented. Experimental results obtained show better force tracking when the neural network is used.

In this paper, the general problem of mapping an n-dimensional polytope Q to an m-dimensional polytope P by P=AQ with n>m is studied. It is well-known that some vertices of Q are mapped to internal points of P, and it is very time consuming to detect them when their number is large. A computationally efficient recursive algorithm based on dimension-growing has been developed which is capable of successively removing the internal points of each intermediate polytope when they are first detected, hence eliminating their further growth into additional internal points in subsequently intermediate polytopes. The method can be applied to compute the robotic manipulability polytope based on the mapping x/spl dot/=Jq/spl dot/. It is shown that substantial reduction in computation time can be achieved for higher DOF redundant robot manipulators. Examples including the 8-DOF ARMII robot are presented and comparisons to the conventional method are made.

This paper presents a simple and computationally efficient scheme for handling a flexible object by two coordinated manipulators. The simulation results for a dual three joints robotic arm system manipulating an object with a single degree of flexible freedom illustrate the ability of the proposed scheme to deform the object to reach a desired shape, and transport the deformed object along a given trajectory. The simulation results also show that it is possible to perform an assembly task, in which the control of contact forces between the object and the fixture is essential to guarantee the success of the assembly.

This paper presents the feasibility study of an adaptive impedance force control application to deburring/milling process. The proposed control regulates desired force on nonhomogeneous workpiece in regardless of unknown location and stiffness of the workpiece. The adaptive technique of minimizing the force error between the desired and the actual force is used to compensate for uncertainties in unknown workpiece. Effect of selecting the impedance damping gain with different workpiece stiffness on force tracking is analytically determined. Analyzing the stability of the proposed adaptive force control yields the bound of the adaptive gain for stability. Simulation studies of deburring/milling process show the excellent force tracking and confirm stability bound for stable force tracking.

A simple adaptive robust force tracking impedance control scheme that has the capability to track a specified desired force is proposed. The new impedance function is realized by the relationship between force error and velocity/acceleration error. The proposed control scheme uses the adaptive technique to compensate for uncertainties in environment location and stiffness. The robot dynamic model uncertainty is compensated by a robust position control algorithm based on time-delayed information. Simulation studies are shown to demonstrate the analysis of the stability and the convergence of the proposed scheme under uncertainties.

A force tracking nonlinear impedance control scheme using a neural network is introduced. Due to the nonlinear characteristics of the proposed control law, stability has been analyzed. The numerical bound of environment position uncertainty for ensuring robot contact with the environment and to perform desired force tracking has been found. Intensive simulation studies with a three link rotary robot manipulator tracking the surface with a desired force are carried out to confirm the boundary solutions of the proposed scheme under no knowledge of environment surface position and environment stiffness.

An impedance function is proposed to achieve accurate force tracking under the presence of uncertainties in robot dynamics and environment models. The new impedance function is formulated on the basis of PID control of the force tracking error which compensates for the unknown environment stiffness and position. The robot dynamics uncertainties are compensated by a simple time-delayed robust control algorithm. Stability and convergence of the control scheme are analyzed. Simulation studies with a three link rotary robot manipulator are shown. Furthermore, experimental results on a PUMA 560 arm are carried out to confirm the proposed impedance controller's performance.

A robust internal force-based impedance control scheme for coordinating manipulators is introduced. Internal force-based impedance control enforces a relationship between the velocity of each manipulator and the internal force on the manipulated objects and requires no knowledge of the object dynamic model. Each manipulator's nonlinear dynamics is compensated by a robust auxiliary controller which is insensitive to robot-model uncertainty and payload variation. The controller is only weakly-dependent on each manipulator's inertia matrix. Stability of the system is analyzed. The scheme is computationally inexpensive and suitable for general-purpose microcomputer implementation. Rigorous experimental investigations are performed and the results presented which validate the proposed concepts.

An internal force-based impedance control scheme for two coordinating robots manipulating a rigid object via palm grasping is introduced. The minimal internal force required to maintain the grasp on the object is computed from the frictional constraints and sensed forces. A closed-form solution to the minimization problem is developed which makes the algorithm suitable for real-time control. The controller uses sensed moments at the palm interface to maintain proper orientation of the palms to achieve maximum surface contact. Each manipulator's nonlinear dynamics is compensated by a robust auxiliary controller which is insensitive to robot-model uncertainty and payload variation. The controller is only weakly-dependent on each manipulator's inertia matrix. Stability of the system is analyzed. Rigorous experimental investigations are are performed and the results presented which validate the proposed concepts.

In this paper, a robust robot force tracking impedance control scheme that uses a neural network as a compensator is proposed. The proposed neural compensator has the capability of making the robot track a specified desired force as well as of compensating for uncertainties in environment location and stiffness, and the uncertainties in robot dynamics. The neural compensator is trained separately for free space motion and contact space motion control using two different training signals. The proposed training signal for force control can be used regardless of the environment profile in order to achieve desired force tracking. Simulation studies with three link rotary robot manipulator are carried out to demonstrate the robustness of the proposed scheme under uncertainties in robot dynamics, environment position and environment stiffness. The results show that excellent force tracking is achieved by the neural network.

An internal force-based impedance control scheme for cooperating manipulators is introduced which controls the motion of the objects being manipulated and the internal force on the objects. The controller enforces a relationship between the velocity of each manipulator and the internal force on the manipulated objects. Each manipulator is directly given the properties of an impedance by the controller; thus, eliminating the gain limitation inherent in the structure of previously proposed schemes. The controller uses the forces sensed at the robot end effectors to compensate for the effects of the objects' dynamics and to compute the internal force using only kinematic relationships. Thus, knowledge of the objects' dynamics is not required. Stability of the system is proven using Lyapunov theory and simulation results are presented validating the proposed concepts. The effect of computational delays in digital control implementations is analyzed vis-a-vis stability and a lower bound derived on the size of the desired manipulator inertia relative to the actual manipulator endpoint inertia. The bound is independent of the sample time.

It is well known that computed torque robot control is subjected to performance degradation due to uncertainties in robot model, and application of neural network (NN) compensation techniques are promising. In this paper we examine the effectiveness of NN as a compensator for the complex problem of Cartesian space control. In particular we examine the differences in system performance when the same NN compensator is applied at different locations in the controller. It is found that using NN to modify the reference trajectory to compensate for model uncertainties is much more effective than the traditional approach of modifying joint torque/force. To facilitate the analysis, a new NN training signal is introduced. The study is extended to non-model based Cartesian control problem. Simulation results are also presented.

Performance of impedance controller for robot force tracking is affected by the uncertainties in the robot model and environment stiffness. The purpose of the paper is to improve the controller robustness by applying the neural network technique to compensate for the uncertainties in the robot model. A novel error signal is proposed for the neural network training. In addition, an algorithm is developed to determine the reference trajectory when the environment stiffness is unknown. Simulations show that highly robust position/force tracking by a three degrees-of-freedom robot can be achieved under large uncertainties.

This paper deals with problems that arise when robot manipulators are used in milling workpieces with nonhomogeneous material. Damages to the cutting tool and workpiece would occur if the cutting speed is set incorrectly with respect to the characteristics of the tool and the hardness of the workpiece material. This concern is heightened when milling materials which are nonhomogenous where the tool drag can vary unpredictably. It is proposed in this paper that we solve the problem by employing a force controlled velocity (FCV) strategy to automatically adjust the cutting speed in response to the force exerted at the tool tip along the cutting path. The idea of FCV is accomplished by using a nonlinear function of both force and elapsed time in place of elapsed time when evaluating the robot trajectory equation. The effectiveness of this approach is confirmed by simulation and experimentation.

The optimal time-dependent receiver (OTDR) is presented and its performance is compared to that of the matched filter receiver. The OTDR, a time-dependent adaptive filter, is shown through simulation to be superior to the matched filter for signals corrupted by cyclostationary interference because it exploits statistical periodicities of the interference.>

The decomposition of forces on an object grasped by multiple manipulators is analyzed using the theory of metric spaces and generalized inverses. In general, the space of forces exerted on the object is nonhomogenous and suitable metrics must be used to decompose the space into motion-inducing and internal force subspaces. A common theoretical framework is proposed in this paper which solves the decomposition problem for rigid, palm-type, and frictional point-contact grasps. New solutions are derived for systems with palm-type and mixed grasps. Previous decompositions for the rigid and frictional point contact grasps are analyzed within the theoretical framework. It is shown that previous solutions in the rigid grasp case are equivalent to the minimization of a norm based on a kinetic energy metric associated with the object. >

This paper proposes a new algorithm, known as the segmentation algorithm, which provides model-based, real-time, whole-arm collision avoidance for telerobotic applications. The work presented is an extension and modification of potential field theory. Novel aspects of the algorithm include the application of a hierarchical segmentation technique to minimize on-line processing and the development of procedures which account for workspace object translation, rotation, and grasping. The segmentation algorithm outputs torques, which, when applied to the control arm, prevent the teleoperator from driving the remote arm into a collision. The teleoperator actually feels workspace objects that are spatially close to the remote arm-an experience known as virtual force-reflection. The segmentation algorithm's performance has been analyzed in terms of its speed and efficiency vis a vis various system parameters, including workspace object distribution, size, and number. Simulation results show that the segmentation algorithm succeeds in providing real-time collision avoidance where less elegant brute force algorithms fail.>

The feasibility and performance of implementing kinematics and inverse dynamics algorithms on a DSP chip for real-time robot arm control is investigated. The algorithms include the following modules: forward and inverse kinematics; Jacobian, inverse Jacobian, and Jacobian derivative term; and Newton-Euler inverse dynamics. These modules are unified under a common coordinate system, and then computationally optimized by eliminating the redundancies among the modules. Further optimization is indicated for the PUMA-like arms. The algorithms are implemented on a TI TMS320C30 DSP chip. It is found that the execution time for the entire set of algorithms is about 0.78 ms for a six-degree-of-freedom robot with a spherical wrist, and is about 0.63 ms for a PUMA-specific arm. The communication time between the host PC and the DSP chip is about 0.376 ms. Thus, it is possible to implement a complete Cartesian controller at a 1000 Hz sampling rate. The algorithms have been successfully tested on a PUMA arm with a PC-based advanced controller. >

A reference force-tracking impedance control system is proposed, consisting of a conventional impedance controller in the inner-loop and a trajectory modifying controller in the outer-loop for force-tracking. The design of the outer-loop is presented and the stability of the two-loop control system is analyzed. A computationally efficient control algorithm for the inner-loop is suggested. Simulation results are presented. The controller is able to achieve excellent position and force-tracking with unknown environment stiffness and the presence of a burr on the tracking surface.>

A special-purpose VLSI array is proposed for performing the computations of the Jacobian, inverse Jacobian, direct kinematics, and numerical integration for the implementation of a joint-space position and velocity command generator for a real-time robot. This integrated approach enhances computational efficiency by reducing the duplicate calculations of these functions and maximizing the parallel/pipelining processing. The array is connected in a systolic manner to form a 6*7 array using 42 basic processing elements (PEs); 6 'circular' PEs, 30 'square' PEs, and 6 'triangular' PEs. A 'square' PE test chip has been designed and fabricated. The test results for this chip show that the array should be able to operate at a step time of 150 ns. For a six-link manipulator with rotary joints, it takes 156 steps, i.e., 23.4 mu s, to complete all the computations in one sampling period.>

The use of a single-camera 3D computer vision system as a position sensor in order to perform robot calibration is examined. A vision feedback scheme, termed vision-guided robot control (VRC), that can improve the accuracy of a robot in an online iterative manner is described. This system demonstrates the advantage that can be achieved by a Cartesian space robot control scheme when end-effector position and orientation are actually sensed instead of calculated from the kinematic equations. The viability of using a vision system for robot calibration is demonstrated by experimentally showing that the accuracy of a robot can be drastically improved. The vision system can also be used to determine the repeatability and accuracy of a robot in a simple, efficient, and quick manner. Experimental work with an IBM electric drive robot and the proposed vision system produced a 97- and 145-fold improvement in the position and orientation accuracy of the robot, respectively.>

The problem of determining collision-free joint space trajectories for redundant robots is considered, and the command-generator approach is used to generate such trajectories. In this approach, a distance objective function is defined and is optimized along the robot joint trajectories through a vector in a null space. Algorithms for implementing this optimization problem are fully developed. It is shown that the proposed collision-free trajectory generation scheme is efficient and practical. Extensive simulation results of a four-link robot example are presented and analyzed.>

It is shown that robot manipulator control can be accomplished using simple decentralized linear time invariant time-delayed joint controllers instead of the complicated computed torque control scheme. This means that all the online computational problems associated with computing robot inverse dynamics can be avoided, and the robot control problem is essentially reduced to that of computing n linear proportional-derivation controls for n joint subsystems. The proposed control technique employs a time-delayed control with a specially designed constant diagonal gain matrix to decouple and linearize the robot joint dynamics so that linear centralized joint control can be achieved. It is shown that the proposed controller is stable, and the value of the special gain matrix can be selected based on a sufficient condition of stability presently developed. However, the establishment of this sufficient condition requires knowledge of the inertial matrix of the robot. It is also shown that the controller is robust in the presence of payload uncertainty. A two-link planar robot is presented to illustrate the controller design procedures and the performance of the controller.>

A joint-trajectory-generation scheme for redundant robots is proposed which uses the redundancy to improve motion performance according to certain objective functions. The objective function can be either analytical or nonanalytical. For nonanalytical objective functions, a least-squares scheme is proposed to estimate the gradient vector. In addition, an approximation scheme is developed to compute the pseudoinverse of the Jacobian. Application of the scheme to a 4-link revolute planar robot manipulator is demonstrated through simulation. Several motion performance criteria are considered and their results analyzed.>

Decentralized control has the advantages of low demand on computer resources and high-speed operation, since the subsystem controllers are implemented in parallel with one another. The authors introduce adaptation to counter the effects of nonlinear and trajectory dependent coupling among the robot joints which cause the independent joint dynamics to vary. They present the results of experimentation with a PUMA 560 arm using a decentralized adaptive control law. The experiments were specially selected for a high degree of intercoupling so that they could investigate how the adaptation mechanism responds.>

An approach is presented for the design of simple robust independent joint controllers which result in linear decoupled joint motions. In this approach each joint is treated as a simple inertial systems plus a disturbance torque representing all the unmodeled dynamics. By means of a very simple procedure, the disturbance is instantly estimated and rejected, thus allowing simple PD control to be used. The controllers are inherently robust. Experimental evaluations of the controller on a PUMA 560 are are performed, and the experimental results are presented.>

A decentralized adaptive control scheme is presented for multilink robot arms. The structure of dynamic and force interactions in the robot arm allows local high gain feedback control to stabilize the overall system, and cause the arm to track reference trajectories to any accuracy desired. The adaptive law is locally stable with an attractive set that is adjustable a-priori using minimal knowledge of robot parameters. Robot manipulators control using decentralized approach is highly desirable in that it facilitates real-time implementation.

Adaptive control has been recognized as an effective approach for mechanical robot manipulator controller design due to the presence of nonlinearties and uncertainties in robot dynamic models. Numerous results addressing different aspects of the control problem have been reported in the literature in recent years. The purpose of this paper is to present an indepth review of these existing developments whereby the robot adaptive control problems and solutions are systematically studied. The essential overall objective is to unify the various design ideas under a common frame work. Within this frame work, the various design techniques are catagorized into three groups; (i) model-reference adaptive control, (ii) Self tuning adaptive control, and (iii) Linear perturbation adaptive control. The design conditions and solutions of each technique are discussed, and the relationships between these techniques are examined. Finally some remarks on the effectiveness and implementation of adaptive controllers are presented.

The main contribution of this paper is the unified treatment of convergence analysis for both LMS and NLMS adaptive algorithms. The following new results are obtained: (i) necessary and sufficient conditions of convergence, (ii) optimal adjustment gains and optimal convergence rates, (iii) interrelationship between LMS and NLMS gains, and (iv) non-stationary algorithm design.