A hybrid GRASP with Rollout for the Bus Driver Scheduling Problem

The Bus Driver Scheduling Problem (BDSP) is a standard Crew Scheduling Problem arising in urban transportation. In this paper, we propose an innovative hybrid metaheuristic that combines GRASP (Greedy Randomized Adaptive Search Procedure) with Rollout. GRASP is a randomized multi-start metaheuristic for combinatorial optimization problems where each iteration leads to the construction of locally optimal solutions, each independent of the others. Rollout is a new heuristic recently proposed in literature by Bertsekas that determines a feasible solution starting from a partial not necessarily feasible solution and applies at each step a base heuristic H. Computational results show that our hybrid heuristic produces better quality solutions when compared with pure Rollout.