An MDP-Based Task Allocation Model for a Class of Uncertain Human-Machine System

摘要

Aimed at a class of small-scale uncertain human-machine systems(SSU-HMS), the problem of the dynamic task allocation is discussed with considering the task uncertainty and the heterogeneity of human operators in this paper. Dynamic task allocation is assigning the right tasks to suitable human operator/automation in real time in order to fully utilize system resources and enhance system performance. In a SSU-HMS, the randomness and uncertainty of the task arrival time and the diversity and variability of task types increase the task uncertainty, which makes task allocation for the system more challenging. In addition, the heterogeneity of individual human operator mainly manifests that they have varying speed, robustness and uneven workload, which make their ability also heterogeneous. Therefore, the potential problem will become more complicated if the heterogeneity is considered. In the work, on the basis of Markov decision process(MDP), a model reflecting the task-operation allocation process of SSU-HMS is established. By iterative method, the optimal task allocation strategy is obtained, which makes full use of system resources and keeps the operators' load at a moderate level. Thus, the overall performance of the system is improved. Finally, the validity of the proposed model is verified by numerical simulations and experimental verification on MATB platform. This paper takes human operator as the research object. In fact, this method is also applicable to the allocation of human and computer resources in human-machine systems.