Gait-based compliant control for snake robots

We present a method of achieving autonomous compliant behavior in snake robots. This behavior allows snake robots to adapt to changes in its environment, e.g. changes in pipe diameter while climbing. To simplify the task of high-level closed-loop control, the controller uses a low-dimensional gait framework that has been previously used for open-loop trajectory generation. We use an extended Kalman filter (EKF) to estimate the gait parameters that best represent the robot's shape, and then choose a control space of parameters relative to that state. Because this controller specifies whole-body motions of the robot, we are able to generate compliant behavior, even though the joints of the robot contain stiff gear ratios with no mechanical compliance or torque sensing.