A Rat-Robot Status Detection Algorithm Based On Machine Vision

Bio-robots take the living biology as the movement carrier, realizing the artificial control of the biological behavior through the Brain-Machine Interface which transmit external control instruction to animal brain area directly. Compared with the traditional robot, Bio-robot has unique advantages in flexibility, concealment, response capacity, energy supply and so on. But now, the control experiment of the Bio-robot still needs to be controlled manually, which limits the application of the Bio-robot in the real environment. It is necessary to realize Bio-robot’s automatic control for perfecting the application scene. And the status detection is an important part of automatic control.This dissertation focuses on the Rat-robot’s status detection in its automatic control experiment. This paper designs detection algorithm and experiment to improve the effect of automatic control by getting higher accuracy of detection results from status information in the top camera image. The main contributions in this dissertation are:(1) This dissertation designs the Rat-robot body position detection algorithm. Improve the detection result by combining the improved Background-Subtraction algorithm and the Gaussian Mixture Model.(2) This dissertation designs the Rat-robot head position detection algorithm. According to the detection result to combine the rat robot region contour feature with corner feature, and using the timing information to optimize the detection results.(3) This dissertation define classification criteria of the robot state according to the information obtained from the status detection. Compare the classify data with manual access data to evaluate the detection results.