Autonomous robotic systems are widely deployed in smart factories and operate in dynamic, uncertain, and human-involved environments that require low-latency and robust fault detection and recovery (FDR). To address this, we propose a novel Goal-oriented Semantic Communication (GSC) framework that minimises the FDR time while maintaining the robotic task success rate requirement. Our GSC framework defines and extracts the 3D scene graph (3D-SG) as the semantic information for FDR via a semantic extractor, and detects faults by monitoring spatial relation changes in the 3D-SG. For fault recovery, we fine-tune a small language model (SLM) and enhance its reasoning and generalization capabilities via knowledge distillation to generate recovery motions for robots with low latency. Extensive simulations demonstrate that our GSC framework reduces the average FDR time by 86.1% while achieving a higher task success rate, compared to the State-of-the-Art (SOTA) frameworks that rely on vision-language models (VLMs) for fault detection and large language models (LLMs) for fault recovery. Project website: https://sites.google.com/view/gscfdr.
Chen et al. (Fri,) studied this question.