miguelgfierro / robot_identification

Humanoid robot identification method using genetic algorithms

Home Page:http://miguelgfierro.com

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Robot identification using genetic algorithms

This is the code to identify the dynamic parameters of a humanoid robot model. It was developed during my PhD thesis. The model is a triple inverted pendulum, with three links and three masses, and the robot is the humanoid HOAP-3.

To identify the system we optimized the triple pendulum parameters minimizing the difference between the theoretical Zero Moment Point (ZMP) and the real ZMP. The ZMP is a criteria for dynamic stability commonly used in humanoids.

The optimization is constrained with respect to the torque limits of the robot. The torques were computed using an inverse dynamics method based on the Lagrange formulation.

The optimization method is a genetic algorithm called Differential Evolution. This algorithm is a good method for multiobjective optimization with constraints.

For more information please visit my blog or my youtube channel

About

Humanoid robot identification method using genetic algorithms

http://miguelgfierro.com


Languages

Language:MATLAB 99.2%Language:M 0.8%