Target Recognition and Tracking with a Rat Robotborne Camera System

Bio-robots, a novel type of robots combing brainmachine interface and living organism, have shown great potential in search and rescue tasks after disasters due to the adaptability and sustained energy efficiency. Nevertheless, image information, the most intuitive source of data, is not effectively utilized in current research. The reason for the lack is the difficulty in the development of camera equipped electric backpack and smooth control of bio-robot to acquire stable and effective images. Regarding this problem, we propose a new method for bio-robot recognizing and tracking target with acquired image information and verify it under experiment. With the new developed camera mounted electric backpack, rat robot system could effectively acquire images. Based on Yolo, we can recognize multiple targets in the images. This research enhances the environmental sensing abilities of bio-robot and promoted it towards practical application in research and rescue tasks.