Project 72: Motion Control of Hexapod Robot based on Reinforcement Learning

Contact Information

Asso. Prof. Yue Gao

Project Description and Objectives

Hexapod robots are highly stable and flexible and are widely used in exploration and rescue missions in complex terrains. However, due to their high-dimensional nonlinear dynamics and complex motion patterns, traditional control methods have certain limitations when dealing with dynamic environments. Reinforcement learning (RL), as a highly adaptive decision-making method, can continuously optimize motion strategies by interacting with the environment and improve the robot's motion performance on complex terrains. This project aims to study and implement hexapod robot motion control based on reinforcement learning. By learning reinforcement learning theory and conducting hexapod robot experiments, students' theoretical knowledge and engineering practice abilities can be comprehensively improved, laying a solid foundation for further in-depth research on robots and reinforcement learning.

Eligibility Requirements

Basic understanding of programming (Python preferred) and deep learning

Main Tasks

Develop a simulation environment tailored for hexapod robots.

Design a reinforcement learning training framework for locomotion tasks.

Transfer the learned locomotion policy to a physical robot to achieve robust walking capability across diverse terrains.

Investigate disturbance perception and adaptive mechanisms to enhance the robot's stability and efficiency under external force disturbances through trained models.

Website

Lab: https://gaoyue.sjtu.edu.cn

School: http://www.seiee.sjtu.edu.cn/