Optimization of Torque Distribution for Enhanced Locomotion in Quadruped Robots

Abstract

Quadruped robots have gained significant attention in recent years due to their versatility and ability to perform challenging tasks. These robots are redundant, possessing more joints than the required degrees of freedom, allowing them to accelerate motion in infinite ways. This redundancy can be exploited to optimize performance, particularly by minimizing joint torque, which also enhances locomotion efficiency and robustness, reduces power consumption, enables handling larger payloads, and even allows for higher speeds of motion. This work resolves actuation redundancy by optimizing the torque for predefined trajectories and gaits. The proposed approach employs advanced redundancy resolution techniques to determine the optimal torque distribution across the joints while adhering to task-specific constraints, such as identifying legs in contact with the ground and preventing leg slippage during ground contact. Additionally, the optimization framework accounts for external loads to improve performance under varying dynamic conditions. The method's effectiveness is validated through simulations using Matlab and Gazebo on the Unitree Go1 robot.