Collaborative Research: Causal Observability and Diagnosis for Human-Robot Interactions Failures in Manufacturing
Senior Personnel · NSF ADMO · PI: Dr. Yisha Xiang
Proposal in preparation
Developing the task “Causal LLM reasoning-trajectory RL” to improve diagnosis of human–robot interaction failures. The proposed work uses reinforcement learning with intermediate causal-reasoning rewards and investigates generative training scenarios to improve generalization across manufacturing tasks.
Planned submission: October 2, 2026