Visual coordination of heterogeneous mobile manipulators

This paper presents work done to control robots of different geometries and capabilities to complete a task neither is capable of independently. The algorithms were tested with a 'cherry picker' experiment that required one large, inaccurate robot to lift and carry a smaller, more accurate mobile robot in order to complete an inspection task by placing its end-effector at a specified distance from a visually-specified target. Both mobile robots are controlled using an uncalibrated visual guidance method. An overview of the algorithm for coordinating the two mobile robots is presented, along with details and results of experiments conducted to measure the accuracy with which the shared task is completed. Results have shown the system to be accurate and robust while requiring very little communication between the two robots.