Optimal viewpoint selection for cooperative visual assistance in multi-robot systems

This paper deals with a monitoring robot system which gives a visual assistance to the working robot. In order to capture an effective scene that gives a proper overview including the working robot and the peripheral objects, we propose an evaluation for the effective composition of the captured image. Based on the evaluation, the monitoring robot autonomously determines the optimal viewpoint according to the motion of the work robot, and moves to the position for the visual assistance. Experimental results show that the scenes from the monitoring robot are better than the best shot of three fixed cameras. Also, the visual assistance is confirmed to be effective for improving remote control of the work robot by our designed user interface.