Adaptive Walking Control of Quadruped Robot Based on Rulkov Neuron Model

Abstract:

To improve the adaptive walking ability of legged robot, a strategy combining the bionic control method and the intelligent optimization algorithm is proposed. The Rulkov neuron model is used to model the central pattern generator (CPG). Based on the CPG model, the single and multi-joint coupling network topology is proposed. The coupling coefficient matrix between CPG units is optimized using the multiobjective genetic algorithm. In this way, the robot’s joints can act correspondingly to timing sequence controlled by the output signals of the CPG network. Finally, the information fusion feedback system and adative walking control strategy are proposed, and a simulation using Webots is implemented on the quadruped robot called GhostDog to experimentally verify it. The experimental results show that the proposed walking control strategy can control the robot to switch walking modes automatically and have certain environmental adaptability.