Rolling-Horizon Task Allocation for Human-Robot Teams with Mental-State Awareness

This paper presents a Mental-State Aware Task Allocation (MSA-TA) framework for human-robot teams in which multiple operators supervise robots executing time-sensitive tasks. Unlike conventional allocation models that treat operators as fixed-capacity resources, the proposed approach includes operator mental state in both the objective function and feasibility constraints. Mental state is represented through a fatigue index and a probabilistic belief over discrete clusters, enabling uncertainty-aware allocation and compatibility with sensing systems such as Brain-Computer Interfaces (BCIs). The problem is solved online through rolling-horizon optimization. Simulation results show that MSA-TA reduces cumulative operator risk exposure and keeps lower fatigue levels than baseline strategies, while preserving competitive task performance.