Based on the Lyapunov theory, a new principle was developed for synthesizing robot tracking control in the presence of model uncertainties. First, a general Lyapunov-like robust tracking concept is presented. It is then used as a basis for the control algorithm derived via a quadratic Lyapunov function constructed using a sliding mode function (based on the output error). Control synthesis is made in task-space, without any need for solving the inverse kinematics problem, i.e. one does not need to inver the Jacobian matrix. It is also shown that the tracking error becomes close to zero in a settling time which is less than a prescribed finite time. Simulation results are incorporated.

Today's industrial machines and manipulators have no capability to learn by experience. Performance and productivity could be greatly enhanced if a machine could modify its operation based on previous actions. This paper presents a learning control scheme that provides the ability for machines to utilize their past experiences. The objective is to have machines mimic the human learning process as closely as possible. A data base is formulated to provide the machine with experience. An optical infrared distance sensor is developed to inform the machine about objects in its working space. A learning control scheme is presented that utilizes the sensory information to enhance machine performance in the next trial. An adaptive scheme is proposed for the modification of learning gain matrices, and is implemented on an industrial robot. Experimental results verify the potentials of the proposed adaptive learning scheme, and illustrate how it can be used for improvement of different manufacturing processes.

The use of a two-arm robot for assembling two objects, with each being held by one arm, is presented. The assembly task is decomposed into an approach phase and an assembly phase. For each phase, we propose a solution for describing the task. For the approach phase, we suggest to describe the task with respect to a mobile reference frame, attached to the end effector of one of the arms. This allows us to take advantage of the redundancy of the system. For the assembly phase, we propose two solutions, both involving some kind of force control. The first one is based upon a position control similar to the one used for the approach phase, with an updating of the reference position through a measurement of the contact forces. The second scheme is derived from a symmetrical hybrid control scheme initially proposed by Uchiyama and Dauchez to control a two-arm robot handling a single rigid object. The main results of this scheme are summarized, and the way of using it for an assembly task is presented. Finally, the experimental setup we have installed to validate our theoretical results is described.

A new approach to motion planning for multiple robots with multi-mode operations is proposed in this paper. Although sharing a common workspace, the robots are assumed to perform periodical tasks independently. The goal is to schedule the motion trajectories of the robots so as to avoid collisions among them. Rather than assigning the robots with different priorities and planning safe motion for only one robot at a time, as is done in most previous studies, an efficient method is developed that can simultaneously generate collision-free motions for the robots with or without priority assignment. Being regarded as a type of job-shop scheduling, the problem is reduced to that of finding a minimaximal path in a disjunctive graph and solved by an extension of the Balas algorithm. The superiority of this approach is demonstrated with various robot operation requirements, including “non-priority”, “with-priority”, and “multicycle” operation modes. Some techniques for speeding up the scheduling process are also presented. The planning results can be described by Gantt charts and executed by a simple “stop-and-go” control scheme. Simulation results on different robot operation modes are also presented to show the feasibility of the proposed approach.

The displacements of a vehicle on a plane can be subject to constraints depending on the nature of the vehicle. One can, for instance, think of the existence of a smallest turning circle for a car.

In this paper our purpose is to show, on a simple example, how such constraints can be handled. We, in fact, consider the case of a vehicle the motions of which consist of a finite sequence of rotations, each rotation being subject to the following constraints.

1) The radius of the circles along which the displacements of the vehicle take place are larger than a critical radius.

2) The centers of the successive rotations are located along a straight line defined by the geometry of the vehicle.

The mathematical analysis of this problem relies on a suitable choice of frames of reference in which the expression of the constraints is particularly simple. It is then shown that, under the above constraints, an arbitrary displacement can always be achieved by three approriate rotations.

Complex tasks that need to be performed through teleoperation led to the development of multifinger robot hands. A dextrous master is a multi-DOF controller which is worn by the operator in order to teleoperate these anthropomorphic hands. Force feedback is very useful when there is interaction with the environment. Providing force feedback to dextrous masters is an area of active research. We outline some of the human factors that influence the design of such masters. Of obvious importance is the hand geometry and the desired number of degrees of freedom. Additional criteria relate to the force perception of the hand. Finally, the man-machine impedance is of importance, since at the man/machine interface there are two impedances acting in series. One is the effective impedance of the human operator holding the master controller, the second is the impedance of the master controller being manipulated.

In this paper, new hybrid control laws for the position control of robotic manipulators are proposed. The proposed control laws are composed of discrete feedforward component and continuous feedback component. The open loop nominal torque about the desired trajectories is taken as the feedforward component, while a modified version of the sliding mode control is taken as the feedback component. For the three proposed control laws, we give sufficient conditions which guarantee the bounded tracking errors in spite of the modeling errors. The existence of the control gains which satisfy these conditions is shown by numerical examples. The computational burden of feedback control is analyzed, which shows that the feedback control can be used in real time digital control. The robustness and the good tracking performance of proposed algorithms are demonstrated by the numerical simulation of a manipulator position control under payloads and parameter uncertainties.

This paper presents a general method to identify the geometric parameters of robots. An algorithm is given to calculate the identifiable geometric parameters. The robot location and the tool location parameters are taken into account. The algorithm is generalized to tree structure robots. The problem of selecting the optimum robot configurations to be used during the identification is discussed and a solution is proposed.

The extraction of contour information from subjects is essential for purposes of grasping and manipulation. We proposed that human haptic exploration of contours, in the absence of vision, would reveal specialized patterns, or “contour exploration procedures,” that are directly related to task goals and intrinsic system capacities. Our general assumptions, method, and initial results were described in Part 1. Part 2 provides an analysis of the relation between contour extraction procedures and processing constraints. These theoretical assumptions are supported by empirical findings, and implications are discussed for issues of importance to robotic exploration and manipulation.

This paper deals with an investigation of the relative importance of robotic characteristics typically associated with nonlinear manipulators. An IBM 7540 SCARA type of robot is used for simulation, and results are presented for decentralized proportional plus derivative control action applied to individual robot joints, and the use of an adaptive computed torque control strategy is illustrated. The influence of variations in payload and robot parameters on trajectory tracking is also shown.

Virtual reference is defined as the signal for a closed-loop System so that the response tracks the desired path perfectly. It is proposed to obtain a virtual reference signal suitable for a manipulator to track a predetermined trajectory using repetitive operations. Thus the algorithm ensures that the difference between the desired and the manipulator trajectories will become negligible as the number of operations approaches infinity if the manipulator is compensated for properly by local feedback. This algorithm uses a dual System to recursively improve the reference signal. Such a System has the advantage of being simple to design. The derivation of this algorithm is based on functional analysis. The effectiveness is confirmed by two experimental results.

An eight-legged walking machine with purely mechanical control of leg motion has been built, based on an old design by Space-General Corporation. The paper describes its mechanical design, and experiments with it on a variety of surfaces.

Actual positions of industrial robot end-effectors differ from those commanded off-line. Consequently, it is then difficult for robots to fulfill certain tasks, such as automated assembly sequences or tasks where high performance accuracy is required. This paper shows that the accuracy of robot performance can be improved by introducing deviation matrices which are functions of many possible error sources. As a first approach, an experiment was carried out where structural elastic deflections, one of the many error sources, of a robot ASEA Irb 6/2 were taken into account. The experiment showed that using the improved model, the positioning accuracy of an ASEA Irb 6/2 robot carrying a weight of 5.6kg was improved from 2.5mm to 0.25mm and the orientation accuracy was improved from 0·45° to 0·3°.

This paper describes the development of assembly strategies based on hybrid force/position control. The assembly strategies developed are confined to the large class of assembly tasks defined as a peg-in-the-hole assembly having a plane of symmetry passing through the axis of insertion.

The basic idea to develop a robotic system for programmable assembly is presented. This system can currently operate controlled manipulators with minor modifications, i.e. if a force/torque sensor and proper software for hybrid control are provided.

The principles of writing assembly strategies are analysed and examples of more complex assembly tasks than the classic round peg-in-the-hole are discussed in detail.

A rigid body dynamical model and control architecture are developed for the closed chain motion of two structurally dissimilar manipulators holding a rigid object in a three-dimensional workspace. The model is first developed in the joint space and then transformed to obtain reduced order equations of motion and a separate set of equations describing the behavior of the generalized contact forces. The problem of solving the joint space and reduced order models for the unknown variables is discussed. A new control architecture consisting of the sum of the outputs of a primary and secondary controller is suggested which, according to the model, decouples the force and position-controlled degrees of freedom during motion of the system. The proposed composite controller enables the designer to develop independent, non-interacting control laws for the force and position control of the complex closed chain system.

The minimum-time motion of robot manipulators is solved by defining a suitable time history for the arm end-effector to traverse. As the planning is performed in the configuration space, the uniqueness of the proposed algorithm emerges from the combination of both cubic and quadratic polynomial splines. Furthermore, the highly efficient time optimisation procedure could be applied to local segments of each joint trajectory, leading to a significant reduction of the travelling time. In addition, the ability to perform a search in the work space is granted, exploiting all possible options for an optimum motion. The method proposed considers all realistic physical limitations inherent in the manipulator design, in addition to any geometric constraints imposed on the path. Simulation programs have been written, and results are reported for the Unimation PUMA 560 robot manipulator.

A robotic manipulator is usually a very complicated system whose dynamics is too complicated time consuming for real-time control. The authors propose general criteria for designing an ideal robot whose dynamics is very much simpler than that of the conventional one. In this paper, dynamic characteristics of an ideally designed robot are investigated. The design guidelines were applied to a 6-degree-of-freedom PUMA 560 robot. Based upon the design concept dynamic equations of the redesigned robot were derived. Utilizing these equations, dynamic characteristics of the redesigned robot are investigated with respect to the computation efforts required, variation of the inertia matrix, joint input torque chracteristics, and couplings between the joints. A detailed comparison study is also made between the redesigned and conventional robots.

In most dynamic adaptive control simulation of robotic manipulators, the Langrange–Euler (L–E) dynamic equations are first piecewise linearized about the desired reference and then discretized and rewritten in a state space form. This makes things very complicated and it is easy to make errors. What is more is that with a different reference this work must be done again. A new simulation scheme – Backward Recursive Self-Tuning Adaptive (BRSTA) – as it will be called, is suggested in this paper for adaptive controller design of robot manipulators. A two degree of freedom robot manipulator is used to verify the scheme in the condition of highly nonlinear and highly coupled system. A one degree of freedom robot manipulator is used for comparing both the forward and backward methods. The main advantages of this scheme include that it can be used for evaluating the self-tuning adaptive control laws and provide the initial process parameters for real-time control. And it is concluded here that the Newton–Euler (N–E) dynamic equations are equally well qualified as the Langrange–Euler (L-E) equations for the simulation of self-tuning adaptive control of robot manipulators.

The use of Multi-Agent Systems as a Distributed AI paradigm for Robotics is the principal aim of our present work. In this paper we consider the needed concepts and a suitable architecture for a set of Agents in order to make it possible for them to cooperate in solving non-trivial tasks.

Agents are sets of different software modules, each one implementing a function required for cooperation. A Monitor, an Acquaintance and Self-knowledge Modules, an Agenda and an Input queue, on the top of each Intelligent System, are fundamental modules that guarantee the process of cooperation, while the overall aim is devoted to the community of cooperative Agents. These Agents, which our testbed concerns, include Vision, Planner, World Model and the Robot itself.

Nonlinear couplings between the various joints of a robotic arm cause trouble in robot control. One possibility to overcome these difficulties is offered by the concept of nonlinear decoupling. The latter leads to independent linear SISO systems, each of them describing the movement of one joint. Thus, an application of control concepts for linear SISO systems is possible. However, at present such decoupling controls are computed from the mathematical model of the arm, the so-called drive equations, whereas actuator dynamics are considered only in a secondary way. In this paper the decoupling problem for robots is investigated by accounting also for the actuator dynamics from the very beginning. This results in decoupling laws requiring a complete state feedback, i.e. not only joint positions and velocities but also the states of the various actuators have to be used. Further, formulas are given which make the computation of those states unsuitable for direct measurements.