Humanoid Robot Modeling and Control (1) - Basic Concepts and Conventions
Humanoid robots, also known as legged robots, have distinct advantages over wheeled robots in rough terrain. However, the advantages of such robots also come with a significant increase in complexity. Although there is now a deep understanding of how to make legged robots dynamically walk and run, further steps are still needed in order to reach higher levels of energy, speed, responsiveness, versatility and robustness. research and efforts. In this blog series, we will explore how to model a humanoid robot, how to perform its stability analysis, how to generate and control its dynamic motion, and the latest trends currently in place to improve its performance. Among them, preventing robots from falling is a major challenge. At the heart of this challenge is the fact that legged robots rely entirely on their contact force with the ground. In this context, the temporality of leg movements appears crucial, since current control strategies mainly focus on the continuous prediction of future movements through model predictive control (MPC), or more specifically on their behavior within limit periods. Orbital stability.
Aspects that will be covered in the future
- FK/IK forward/inverse kinematics
- FD/ID forward/inverse kinematics
- MPC/NMPC linear/nonlinear model predictive control
Aspects that may be covered in future columns
- RL based control Control based on reinforcement learning
- RT1& RT2 etc.