Data-driven control of planar snake robot locomotion

Abstract

A direct data-driven strategy for snake-robot locomotion control is proposed in this paper. The approach leads to a time-varying state feedback controller with robustness guarantees. Instead of relying on exact model knowledge - which is often not available in practice - the proposed control strategy requires only input-state data collected during offline experiments. The efficacy of the proposed strategy is demonstrated via simulations. Notably, by using data to compensate for inaccurate models, the proposed control strategy can lead to significant improvements in closed-loop performance compared to existing (model-based) control strategies, while also eliminating the need for manual tuning of control parameters.