From neuron to motion: hybrid CPG-muscle modeling for biologically plausible robotics

this study presents a bio-inspired locomotion control architecture that integrates a half-center model central pattern generator (CPG) based on spiking neurons with a biophysically grounded muscle contraction model. The CPG operates through mutually inhibiting half-centers generating antiphasic activity to control antagonistic muscles. A key innovation involves the direct use of motor neuron spiking activity to drive the Wexler muscle contraction model, eliminating the need for signal averaging and ensuring physiologically realistic force dynamics. Simulation results demonstrate high stability in force generation by antagonistic muscle pairs (variation ≤3%) and the production of smooth rhythmic patterns for controlling snake-like robot servomotors. The proposed approach significantly enhances control smoothness and adaptability compared to traditional methods, offering new prospects for developing energy-efficient and highly maneuverable bionic robotic systems.