October 10, 2025Open Access

Learning a Unified Policy for Position and Force Control in Legged Loco-Manipulation

View Full Paper

Ask AI

Bookmark

A unified policy achieves approximately 39.5% higher success rates in manipulation tasks compared to position-control policies.
Simulations incorporated diverse position and force commands to train the robot without force sensors, enhancing adaptability.
Reinforcement learning was employed to learn force compensation through position and velocity adjustments from historical states.
Extensive validation on quadrupedal and humanoid robots confirms the policy's robustness across various contact-rich scenarios.

Discussion

No takes yet. Share an insight, caveat, or question.

Proposes a reinforcement learning policy for joint force and position control in robots, indicating enhanced manipulation capabilities.

Zhi et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark