Adaptive task allocation and scheduling for human–robot collaborative construction under dynamic uncertainty

Human–robot collaborative (HRC) construction requires task allocation and scheduling that can accommodate changing site conditions while balancing duration and human fatigue. This paper examines how task representation, task–agent alignment, and rescheduling can jointly balance these objectives under dynamic uncertainty. A framework integrating directed acyclic graph generation from work breakdown structure, non-additive alignment, bi-objective optimization, and event-driven local repair is developed and evaluated using an interior finishing case with static baselines and repeated disruption experiments. Static optimization yielded distinct Pareto endpoints with a minimum makespan of 52.43 h and a minimum fatigue index of 0.89, while dynamic repair achieved the lowest mean fatigue index among the baselines (0.903) with a makespan of 58.04 h. These results help construction planners compare scheduling efficiency with human fatigue and adapt schedule quickly during disruptions. They motivate multi-trade field validation using worker-specific biomechanical measurements, site disruption records, and richer uncertainty and safety models.