Humanoids adjust end-effector and body stiffness for safe interaction
CEER2: Directional and Tunable End-Effector and Root Compliance for Humanoid Loco-Manipulation
Robotics
Summary
When humanoid robots touch people or objects, they need to react carefully so they don’t cause harm but still complete their tasks. The authors created a system that lets the robot soften or stiffen different parts of its body and hands in specific directions, depending on the situation. This system uses machine learning to decide how to respond based on sensing forces it feels. They tested it in both computer simulations and on a real robot, showing it can smoothly handle tasks like lifting and carrying things together with a person.
What this means in practice
- •For robotics engineers: Build humanoid robots that safely and precisely adjust stiffness during physical interaction with humans and environments.
- •For automation integrators: Implement advanced compliance control for robots collaborating on carrying and manipulating objects with people.$Commercial implications: Enables development of cooperative humanoid robots for logistics and manufacturing that adapt stiffness dynamically for safe teamwork.
Authors
Xinyuan Luo, Chunyuan Yang, Boyuan Chen, Xianyi Cheng
Abstract
Humanoids are increasingly capable of tracking complex whole-body motions, but physical interaction introduces a different challenge. When a robot makes contact with a person or the environment, it needs to respond to external forces while preserving the motion needed for the task. This response can vary across directions in the end-effectors and on the body. For example, an end effector may need to accommodate contact force in one direction while maintaining motion accuracy in another, while the robot body may resist an external force or move with it. We present a compliance framework for humanoid loco-manipulation that combines directional and tunable end-effector (EE) compliance with selectable root compliance for external force rejection or force following. A hierarchical reinforcement learning controller modulates a fixed whole-body tracking policy through high-level EE and root commands, while interaction forces are estimated from proprioceptive history. Our simulation and real-world experiments on a humanoid demonstrate directional stiffness control, online stiffness adjustment, distinct root compliance, compliant manipulation, and collaborative carrying.