Whole-Body Dynamics for Humanoid Robot Fall Protection Trajectory Generation with Wall Support

Area of Science:

• Robotics
• Control Systems
• Humanoid Robot Dynamics

Background:

• Humanoid robots operating in human environments face challenges with stability and fall prevention.
• Utilizing environmental objects offers a potential solution for maintaining robot balance.

Purpose of the Study:

• To develop and validate a whole-body motion control strategy for humanoid robots to leverage environmental objects for stability.
• To reduce impact forces and ensure safe landings when interacting with obstacles like walls.

Main Methods:

• Established robot state equations using a variable height-inverted pendulum model.
• Implemented online trajectory optimization with model predictive control and the distributed polygon method for arm postures.
• Developed a multi-objective cost function for whole-body motion control, incorporating dynamics, center of mass, end effector, friction, and center of pressure constraints.

Main Results:

• Simulations showed a reduction in maximum acceleration upon wall impact to 69.1 m/s², significantly decreasing impact force compared to free fall.
• Actual experiments demonstrated stable wall landings with impact forces within acceptable limits for the robot.

Conclusions:

• The proposed whole-body motion control method effectively enables humanoid robots to utilize environmental objects for stable landings.
• The approach enhances robot safety and robustness in dynamic human environments by mitigating impact forces.