CONTROL OF COOPERATIVE DUAL-ARM MOBILE ROBOTS IN A VISION-BASED INTELLIGENT SPACE

摘要

This paper presents a seemingly novel approach to accomplish cooperative tasks with two dual-arm mobile robots in a vision-based intelligent space. Specifically, the cooperative task considered is to grasp and move a box object to a target position with a ceiling-mounted IP camera. To grasp the box, the desired poses of the dual arms are determined based on the poses of the mobile robots and the box. The movement of the box to its desired position is then performed by two decentralized visual servo controllers for the two mobile robots. In particular, the master robot is controlled to reach its desired pose while the slave robot is driven to maintain the grasping employing compliance control and to keep relative pose to the box using real-time visual servoing requiring no communication with the master robot. The proposed approach has been successfully validated by experimenting with two Dr Robot i90 robots equipped with 6-DOF dual arms. The system can be further applied to cooperative tasks in a large workspace equipped with a set of fix-mounted vision sensors.