Attractoring-based locomotion for hexapods

Generating robust and adaptable legged locomotion with minimal control architecture remains an open challenge in bio-inspired robotics. Existing central pattern generator (CPG) approaches often rely on multi-neuron oscillators, asymmetrical network structures, abstract phase oscillators, or task-specific tuning to produce stable gaits. Here, we address this problem by introducing a minimal sensorimotor control framework based on single-neuron CPGs with proprioceptive feedback. Through stability analysis and physical experiments, we show that fully symmetric coupling between single-neuron units is sufficient to generate self-organized tripod-type gaits, enable reliable gait switching via single-pulse kick control, and sustain locomotion even under leg failure. In the strong-attractoring limit, coordinated locomotion emerges without intrinsic neural oscillations, driven solely by sensory feedback. The same framework, without parameter changes, also produces coordinated quadruped locomotion, illustrating its generality. This demonstrates that complex and robust locomotor patterns can arise from extremely simple decentralized mechanisms. Our results contribute to the search for generative principles underlying locomotion and provide a lightweight, extensible basis for bio-inspired control across diverse robotic platforms.