Event-Based Target Tracking Control for a Snake Robot Using a Dynamic Vision Sensor

Dynamic Vision Sensor (DVS) is a promising neuromorphic vision sensor for autonomous locomotion control of mobile robots, as the DVS acquires visual information by mimicking retina to sense and encode the world as neural signals. In this paper, we present an autonomous target detecting and tracking control approach for a snake-like robot with a monocular DVS. By using Hough transform based on the Spiking Neural Network (SNN), the target pole is detected as two parallel lines from the event-based visual input. Then a depth estimation method based on the pose and motion of the robot is proposed. Furthermore, by combining the periodic motion feature of the snake-like robot, an adaptive tracking method based on the estimated depth information is introduced. Experiments are conducted on a snake-like robot to demonstrate the practicality and accuracy of our proposed method to track a target pole dynamically with a monocular DVS.