Squat Motion of a Humanoid Robot Using Three-Particle Model Predictive Control and Whole-Body Control

Area of Science:

• Robotics
• Control Systems
• Humanoid Robot Motion

Background:

• Squatting is a fundamental movement for humanoid robots, essential for tasks like robot commissioning, service industries, and dynamic actions such as jumping.
• Precise and continuous squatting is critical for advancing humanoid robot capabilities and applications.

Purpose of the Study:

• To develop and evaluate a novel control algorithm for achieving accurate and continuous squatting motions in humanoid robots.
• To improve trajectory tracking and mitigate issues like excessive knee torque spikes during squatting.

Main Methods:

• A three-particle model predictive control (TP-MPC) approach was combined with weight-based whole-body control (WBC).
• Humanoid robot dynamics were simplified into three particles (arms, legs, torso) for TP-MPC optimization.
• TP-MPC generated optimized reference trajectories, which were then followed by WBC.

Main Results:

• The combined TP-MPC and WBC algorithm demonstrated superior trajectory tracking accuracy compared to WBC alone.
• The proposed method effectively reduced excessive knee torque spikes observed with WBC during squatting.
• The algorithm proved to be computationally efficient, operating at a frequency of 100 Hz.

Conclusions:

• The integration of TP-MPC with WBC offers a significant advancement for controlling humanoid robot squatting motions.
• This approach enhances precision, reduces undesirable torque spikes, and maintains computational efficiency.
• The findings contribute to the development of more capable and robust humanoid robots for diverse applications.