Optimising stochastic task allocation and scheduling plans for mission workers subject to learning-forgetting, fatigue-recovery, and stress-recovery effects

This study addresses the stochastic bi-objective task allocation and scheduling problems for mission workers subject to the complex (and joint) effect of learning-forgetting, fatigue-recovery, and stress-recovery processes. The mission consists of work-rest cycles with different types of tasks. Some task types are repetitive, not necessarily back-to-back, with some predecessors for other task types. The workers are multi-skilled, whose experience, learning, fatigue, and stress levels are updated, affecting their performance dynamically. A Markov Decision Process (MDP) is applied to formulate this stochastic problem, considering speed and accuracy as two measures of workforce performance. A decision could be "to repeat a task of a certain type" or "take a rest break till another worker becomes available". The developed MDP model finds the optimal task allocation and workbreak schedule for workers by minimising the sum of the tasks' completion times and maximising their quality. Completion time is a continuous variable, and the quality index is a binary random variable, i.e., high or moderate, having a continuous probability of occurrence. The model is of a general form with a potential application in similar settings. The paper used the Sequential Greedy Assignment (SGA) and the Monte-Carlo Tree Search (MCTS) to solve the problem with their results compared. Numerical results with those from a sensitivity analysis are discussed.