Achieving a Cooperative Behavior in a Dual-Arm Robot System via a Modular Control Structure • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • Fabrizio Caccavale, * Ciro Natale, Bruno Siciliano, and Luigi Villani PRISMA Lab Dipartimento di Informatica e Sistemistica Universit` a degli Studi di Napoli Federico II Napoli, Italy e-mail: caccavale@unina.it Received 16 August 2001; accepted 16 August 2001 In this paper the problem of achieving a cooperative behavior in a dual-arm robot system is addressed. A control strategy is conceived in which one robot is position controlled and devoted to task execution, whereas a suitable compliance is conferred to the end effector of the other robot to cope with unavoidable misalignment between the two arms and task planning inaccuracies. A modular control structure is adopted that allows the selection of the proper operating mode for each robot, depending on the task requirements. The proposed approach is experimentally tested in two different tasks involving the two robots in the laboratory setup. First, a parts-mating task of peg-in-hole type is executed; the robot carrying the peg is position controlled, whereas the robot holding the hollow part is controlled to behave as a mechanical impedance. Then, a pressure-forming task is executed, in which a disk-shaped tool is required to align with a flat surface while exerting a desired pressure; in this case, the robot carrying the disk is position controlled, whereas the robot holding the surface is force controlled. © 2001 John Wiley & Sons, Inc. 1. INTRODUCTION The adoption of multiarm systems in lieu of a sin- gle robot arm is crucial for the execution of many robotic tasks—for example, parts mating, and trans- portation of heavy or large objects. In fact, this is just ∗ To whom all correspondence should be addressed. the behavior commonly adopted by humans for the execution of complex manipulation tasks. Most robotic tasks involving interaction between two parts cannot be successfully carried out by relying on a purely positional control strategy. For example, when assembly of rigid parts is of concern, even small path planning errors and/or control position inaccu- racies may induce undesirable large values of contact Journal of Robotic Systems 18(12), 691–699 (2001) © 2001 by John Wiley & Sons, Inc.