This special issue is concerned with the intelligent control of robots in the performance of manipulation tasks. First we will try to define what we mean by intelligent control and by manipulation tasks.

Unfortunately, intelligent control is a somewhat imprecise term, just like the word control itself, so that its meaning can vary widely depending on the application and on the writer's scientific discipline. In general terms, we can say that intelligent control implies some degree of autonomy in performing a task while accommodating uncertainty. In the control of sensor-based robot systems, we could characterize an intelligent controller as one with the following capabilities;

Learning control is a new approach to the probelm of skill refinement for robotic manipulators. It is considered to be a mathematical model of motor program learning for skilled motions in the central nervous system.

This paper proposes a class of learning control algorithms for improving operations of the robot arm under a geometrical end-point constraint at the next trial on the basis of the previous operation data. The command input torque is updated by a linear modification of present joint velocity errors deviated from the desired velocity trajectory in addition to the previous input. It is shown that motion trajectories approach an e-neighborhood of the desired one in the sense of squared integral norm provided the local feedback loop consists of both position and velocity feedbacks plus a feedback term of the error force vector between the reactive force and desired force on the end-point constrained surface. It is explored that various passivity properties of residual error dynamics of the manipulator play a crucial role in the proof of uniform boundedness and convergence of position and velocity trajectories.

It has been experimentally verified that the jerk of the desired trajectory adversely affects the performance of the tracking control algorithms for robotic manipulators. In this paper, we investigate the reasons behind this effect, and state the trajectory planning problem as an optimization problem that minimizes a norm of joint jerk over a prespecified Cartesian space trajectory. The necessary conditions are derived and a numerical algorithm is presented.

A method is described to convert information available at manipulator programming level into trajectories which are suitable for tracking by a servo control system. This process generates trajectories in real time which comply with general dynamic and kinematic constraints. Tracking accuracy will depend mainly on the acceleration demand of the nominal trajectory setpoints - the actuator output demands, in particular, must remain bounded. Our scheme takes into consideration at the trajectory computation level the dynamics of the underlying system, dynamically available information acquired through sensors, and various types of constraints, such as manipulators. It has been developed in the context of a multi-manipulator programming and control system called Kali and developed at McGill University.

The Flight Telerobotic Servicer (FTS) is a robotic device which will be used to build and maintain Space Station Freedom. The FTS is expected to evolve from its initial capability of teleoperation toward greater autonomy by taking advantage of advances in technology as they become available. In order to support this evolution, NASA has chosen the NASA/NIST Standard Reference model for Telerobot Control System Architecture (NASREM) as the FTS functional architecture. As a result of the definition of generic interfaces in NASREM, the system can be modified without major impact. Consequently, different approaches to solve a problem can be tested easily. This paper describes the implementation of NASREM in the NIST laboratory. The approach is to build a flexible testbed to enhance research in robot control, computer vision, and related areas. To illustrate the real-time aspects of the implementation, a sensory interactive motion control experiment will be described.

Assembly problems require that a robot with fewer actuated degrees of freedom manipulate an environment containing a greater number of unactuated degrees of freedom. From the perspective of control theory, these problems hold considerable interest because they are characterized by the presence of non-holonomic constraints that preclude the possibility of feedback stabilization. In this sense they necessitate the introduction of a hierarchical controller. This paper explores these issues in the simple instance when all of the pieces to be assembled are constrained to lie on a line. A hierarchical controller is devised for this problem and is shown to be correct: the closed loop system achieves any desired final assembly from all initial configurations that lie in its connected component in configuration space; the generated sequence of motions never causes collisions between two pieces. Further examination of this approach interprets the controller's mediation of conflicting subgoals as promoting an M-player game amongst the pieces to be assembled.

The work demonstrates a new approach to design of a level of intelligent control of robotic systems. The analytical model results in operational decisions. The structure of these decisions make them readily available to be implemented as an expert system. The approach is applied to a case study of mobile supervisory robots. The model presented here was motivated by manufacturing robotic systems and a type of autonomous robots that collect information at different sites for safety and other control purposes. The robot actions are directly affected by the obta~ned data. At each site the amount of available information (and thus the correctness of the robot decision) can be increased if the robot keeps collecting data at that site for a longer period of t~me. This means a delay in reacting and in reaching next site and accordingly, a decrease in the general amount of robot's information on the whole system.

The method of finding an economic amount of information collected by a robot at each site is based on the theory of controlled discrete event stochastic systems developed in our earlier works. This theory combines he basic concepts of discrete event control extended to stochastic systems with some aspects of information economics.

We examine the properties of an integrated, intelligent controller for multifingered manipulators. The discussion addresses the dynamic constraints on computation and resource allocation in systems which must continually adjust objectives and the set of behaviors employed to attain them. We present a means of specifying general force domain tasks and include techniques for reflexive, sensor-based control and for optimal combinatoric grasp synthesis as alternative grasp formation mechanisms. Examples are presented to illustrate the acquisition of information, the incremental refinement of models, the use of uncertain information in reflexive grasp formation, and the use of complete geometric information in optimal grasping.

This paper presents a qualitative theoretical formulation for synthesis and analysis of multiple contact dexterous manipulation of an object, using a robot hand. The motivation for a qualitative theory is to build a formalisation of ‘human-like’ common-sense reasoning in robotic manipulation. Using this formalisation, a robot hand can perform finger-tip manipulative movements by analysing the physical laws that govern the robot hand, the object, and their interaction. Traditionally, such analysis have been framed in quantitative terms leading to mathematical systems which become intractable very quickly. Also, quantitative synthesis and analysis, often demand an accurate specification of the parameters in the universe of discourse, which is almost impossible to provide. The qualitative approach inherently encounters both these problems successfully.

The qualitative theory is presented in three developmental stages. A qualitative framework of spatial information in the context of dexterous manipulation has been provided. Qualitative models of an object configuration and transformations in them that occur during a manipulation process, have been developed. Finally, the development of a ‘quasi-static’ qualitative framework of a dexterous manipulation process that performs the desired object transformation, has been presented.

This paper presents an advanced control system for an active compliant device. This device, a manipulator-gripper, was designed to achieve stable grasp of objects with various shapes and to impart compliant fine motions to the grasped object. In the control system of this end-effector, we introduced autonomous reasoning capabilities. Fine motion strategies, needed for mating or grasping, use inductive learning from experiments to achieve uncertainty and error recovery. An overview of the articulated gripper's structure is provided for a better understanding of the programming environment we propose. For solving the problem of synthesis programs for fine motion planning we introduce declarative programming facilities in the controller through a time-sensitive mini-prolog. The paper gives some details on the implementation of this mini-prolog. We develop a heuristic procedure to obtain an implicit local model of contacts in complex assembly tasks. Finally, a specific example of this approach – a peg-in-hole operation– –is outlined.