Logo image
Entropy-Weighted Simulated Annealing optimisation of Human-simulated Multi-mode PD-PI Control for Biped Robots
Conference proceeding

Entropy-Weighted Simulated Annealing optimisation of Human-simulated Multi-mode PD-PI Control for Biped Robots

Xingyang Liu, Ferrante Neri, Daniel Cyrus, Haina Rong and Gexiang Zhang
2024 IEEE Conference on Artificial Intelligence (CAI), pp.524-529
25/06/2024

Abstract

Accuracy biped robots Entropy entropy weight method Human-simulated intelligent control Legged locomotion multi-mode control PI control Search problems Simulated annealing Stability analysis
Facing the challenges of biped robot walking control, this study introduces an innovative multi-mode PD-PI (Proportional-Differential Proportional-Integral) controller. The design of this controller is inspired by human-simulated intelligent control, aiming to enhance the accuracy and stability of joint motion. The setting of the parameters of the controller results in a complex multi-objective optimisation problem. To effectively tune the controller, this study incorporates a combination of entropy weight method and simulated annealing algorithm. Simulation experiments performed on the traditional PD controller and on the proposed PD-PI demonstrate that the optimised PD-PI controller significantly improves upon the traditional PD performance in terms of control precision and joint stability. These improvements highlight its potential advantages in advancing the gait of robots.

Metrics

1 Record Views

Details

Logo image

Usage Policy