Real-Time Remote Control Framework for Humanoid Robots Based on Motion Capture

In hazardous working environments, remote operation methods are essential to ensuring personnel safety and mitigating potential risks. To address this challenge, a remotely operated robotic system is proposed to perform complex tasks, reducing risks to human operators. This paper introduces a lightweight real-time control framework that balances simplicity and effectiveness in robotic operations. Human motion data are captured using an Xsens motion capture system on a sender computer and transmitted via User Datagram Protocol (UDP) to a receiver computer. The Robot Operating System (ROS) is utilized on the receiver computer to process the human body data by removing unnecessary torso information, retaining key joint data, and mapping it to the joints of the humanoid robot. Additionally, the joint data is smoothed to enhance the movements of the robot, making its locomotion smoother and more natural. Finally, the feasibility of the framework is validated through MuJoCo simulations and real-world experiments on the Unitree G1 humanoid robot.