CPG's Parameters and Topology Co-Evolution for Walking Control of Biped Robots

• 作 者: XIAOHui;CHENQijun;LIUChengju;
• 机 构: Department of Electronic and Information Engineering, Institute of Control Theory and Control Engine
• 出 处: 2013年中国信息通信研究新进展研讨会暨第五届数字媒体技术专业建设研讨会
• 发表日期: 2013年期

　　In this paper, we focus on the bipedal walking that can be achieved by using a bio-inspired controller based on central pattern generator (CPG).The challenge of this work is to determine the topology of neural network and the setting of the parameters.For this, by using neural oscillators to generate joint position control signals directly, a proper distributed oscillator network which consists of a body network and a leg network is constructed.To realize stable biped walking, the sensory feedback loop is designed and the parameters of the system are evolved by multi-objective genetic algorithm (MOGA).The presented control method is validated through an ODE-based physically simulated environment.Under the controller, NAO can realize basic walking pattern, which demonstrate the effectiveness of the presented bio-inspired control method.