POMA: Propagation-Based Obstacle Negotiation Control for Multi-segmented Robot Adaptation

This paper presents Propagation-based Obstacle negotiation control for Multi-segmented robot Adaptation (POMA) in complex environments. POMA integrates three neural control mechanisms: central pattern generator (CPG)-based leg control for generating gaits, horizontal body control for avoiding high obstacles, and vertical body control for climbing ramps and small obstacles. We validated the performance of POMA in a physical simulation. Our experimental results show that POMA enables a bio-inspired multi-segmented, legged robot to adaptively and successfully navigate through a maze with up and down ramps without a map of the environment. This demonstrates the effectiveness of integrating different neural control mechanisms for multi-segmented robots to deal with complex environments.