🤖 AI Summary
To address the challenges of prolonged task duration, high error rates, excessive operator cognitive load, and degraded situation awareness in deep-space teleoperation, this work proposes a closed-loop teleoperation system for lunar surface robot assembly and fault response. The system introduces the first lightweight, dynamic digital twin framework integrated into a teleoperation architecture. It combines ROS2-based real-time communication, Unity3D physics simulation, and a latency-adaptive synchronization algorithm to enable millisecond-level bidirectional state mapping and online fault reasoning between ground simulation and lunar robots—designed for low-bandwidth deep-space links. Innovatively, it integrates multimodal human–robot interfaces with a physics-informed motion prediction mechanism. In simulated lunar environments, the system achieves 98.7% accuracy in assembly action prediction, an average fault response time of 4.2 seconds, and an end-to-end task success rate of 93.5%.