DYNAMIC TASK MAPPING FOR REAL-TIME CONTROLLER OF DISTRIBUTED COOPERATIVE ROBOT SYSTEMS Tim Lueth, Thomas Laengle, and Jochen Heinzman Institute for Real-Time Computer Systems and Robotics IPR University of Karlsruhe, D-76128 Karlsruhe F.R. Germany, email: t.lueth@ieee.org Abstract: In the KACORs (Karlsruhe Cooperative Robots) project, several small mobile manipulators are used to support the bigger mobile two-arm robot KAMRO (Karlsruhe Autonomous Mobile Robot) during assembly tasks. All robots are programmable on task level and the task description is transformed into different sequences of elementary operations (skills, behaviors) which are executed by the real-time robot controller. Each elementary operation consists of a set of control loops that must be executed in real-time in the robot controller of the small robots. The paper describes the general framework for cooperative task execution, the architecture of the robot controller, and the problem of managing the on-line mapping and downloading of the real-time control loops to the robot. Keywords: Control theory, Distributed Artificial Intelligence, Decentralised Systems 1. INTRODUCTION Up to now, intelligent robot control is formulated fuzzy, and seems to be more an art than a theory. Many authors have already tried to solve this problem and offered similar and different theories for the intelligent control of robot systems (Albus et al. 1981; Albus, 1991; Antsaklis, 1994; Braitenberg, 1984; Brooks, 1991; Hayes-Roth, 1988; Meystel, 1989; Musliner et al. 1993; Valavanis and Saridis, 1992; Walter, 1961; Wiener, 1948). It is not tried here to introduce a further variant, but to describe the already presented concepts in an easily understandable framework (Lueth, 1995; Lueth et al., 1995). In this paper, the concepts of the integrated control theory are presented by use of several projects at the University of Karlsruhe. In the KACORs (Karlsruhe Cooperative Robots) project, several small mobile manipulators are used to support the bigger mobile two-arm robot KAMRO (Karlsruhe Autonomous Mobile Robot) during assembly tasks. Using the common terms, which are common in robotics, the KAMRO system is driven by a hierarchical intelligent control system, and the KACOR systems are driven by a behavior based approach. Nevertheless, it is possible to describe the control systems of both robot types, which continuously use sensor information for unsupervised task execution, by a network of control loops that run parallel to each other. The control loops are redundant and independently processing the same sensor information, and they control the same actuators to concurrently achieve several conditions. Therefore, the intelligent controller can be described as a distributed control system which has to ensure a minimal response time to an environment change. This is the key to achieve robust behavior on control level.