Quadruped Robot Leg Optimization Based on a Multi-Objective Genetic Algorithm

To improve the speed of a quadruped robot, a novel electric-drive robot leg based on a double four-bar mechanism is proposed. This scheme avoids the need for constant alternation between the acceleration and deceleration of the motors and may thus achieve a higher rotation speed. However, the dimensional parameters of the double four-bar mechanism have a significant impact on the dynamic performance of the robot. To determine the optimal dimensions, a kinematic and dynamic analysis of the proposed robot leg is carried out, and the functional relations between the joint motors peak torque, peak angular velocity and total energy consumption in a single gait cycle and the dimensional parameters are determined. In this way, a multi-objective optimization model for the leg design is constructed, and the gamultiobj algorithm is applied to obtain the Pareto optimal solution set of the optimization model. By determining the weight of each objective function based on a combination weighting method, the optimal dimensions of the leg are obtained from the Pareto optimal solution. Finally, a virtual prototype of the robot leg is built via ADAMS to conduct a walking simulation, and the physical prototype is installed on a self-developed test bench of one-legged movement to conduct a no-load walking experiment. The results of the theoretical calculations and simulation are well coincident with those of the tests, which verify the correctness and validity of the proposed algorithm.