Neurally Controlled Steering for Collision-Free Behavior of a Snake Robot

Abstract

Biologically inspired snake robots have been widely studied for their various motion patterns. Most research has focused on the design of a controller for a given motion pattern. However, relatively limited work appears to have been done on the design of a controller for self-adaptive locomotion. In this brief, we add sensory inputs to a control system in order to investigate collision avoidance in a snake robot using a neural controller based on central pattern generator. From an analysis of the steering mechanism during serpentine locomotion, we derive a mathematical model of the joint configuration and the steering angle. In a neural oscillator network, steering control can be achieved via the proposed amplitude modulation method by modulating the neural oscillation parameters. A head-navigated motion pattern is employed to allow the range sensors to accurately detect obstacles for collision avoidance. Through the head-navigated locomotion, the head of the snake robot can be controlled to keep the orientation the same as the motion direction. The proposed control method is experimentally verified by application to the SR-I snake robot.