Hierarchical RL-Guided Large-scale Navigation of a Snake Robot

TL;DR

This paper introduces a four-layer hierarchical control scheme combining reinforcement learning and central pattern generators to enable a snake robot to navigate large, complex environments adaptively and efficiently.

Contribution

The work presents a novel hierarchical control framework that integrates RL and CPGs for large-scale, adaptive snake robot navigation in unstructured environments.

Findings

Learned to navigate complex mazes within hours

Can be deployed to new environments in zero-shot

Effective in large-scale, unstructured settings

Abstract

Classical snake robot control leverages mimicking snake-like gaits tuned for specific environments. However, to operate adaptively in unstructured environments, gait generation must be dynamically scheduled. In this work, we present a four-layer hierarchical control scheme to enable the snake robot to navigate freely in large-scale environments. The proposed model decomposes navigation into global planning, local planning, gait generation, and gait tracking. Using reinforcement learning (RL) and a central pattern generator (CPG), our method learns to navigate in complex mazes within hours and can be directly deployed to arbitrary new environments in a zero-shot fashion. We use the high-fidelity model of Northeastern's slithering robot COBRA to test the effectiveness of the proposed hierarchical control approach.