Optimal Trajectory Planning and Feedback Control of Lateral Undulation in Snake-Like Robots

摘要

Although the lateral undulation gait has received much attention in the field of biologically-inspired snake robotics, tools that support its practical use as a locomotive method remain scarce. Using a recently developed dynamic model for lateral undulation, this paper describes an optimal control trajectory synthesis strategy and a closed-loop feedback control approach to trajectory tracking. Enabling the two is an empirical control-to-action mapping whose input is a gait parameter vector and whose output is the averaged steady-behavior body velocity of the system, approximating lateral undulation with a kinematic model. Direct collocation-based trajectory optimization with the kinematic model along with Lp-norm representations of the robot body and obstacles for collision avoidance, produce optimal, dynamically-feasible paths for navigating to a goal position. During path tracking, the feedback control corrections are mapped back to appropriate gait parameter commands using the generated map. Application of the trajectory planning and feedback controlled path following strategy in simulation demonstrates successful traversal of various obstacle arrangements, even in the presence of modeling approximation errors and unmodeled gait transition dynamics.