Design of trot gait parameters planning system for parallel quadruped robot based on virtual model controller and fuzzy neural network

The capability to achieve fast motion in varying road conditions is a crucial research aspect in the dynamic control of quadruped robot. In this study, a gait parameters planning system for quadruped robot based on virtual model controller (VMC) and fuzzy neural network controller (FNNC) is proposed. According to the expert knowledge, the FNNC is designed to help optimize the parameters in the central pattern generator and virtual model controller (CPG-VMC). This affect the performance of the fast motion indicated by the attitude plantar force and a weight adaptive law is designed and implemented to improve the capability of traversing unprecedented road conditions. To better analyze controller efficiency, the concept called cost of transport (CoT) is introduced to serve as the evaluation criteria for the performance of controller. Both the simulation and prototype test are implemented to validate the effect of the proposed method. Experimental results show that the FNNC-based gait parameters planning system can accurately detect the flaws in the parameters, help adjusting the parameters in real-time regarding the different road conditions, and reducing the CoT and the vibration.