Fuzzy linear assignment problem: an approach to vehicle fleet deployment

This paper proposes and examines a new approach using fuzzy logic to vehicle fleet deployment. Fleet deployment is viewed as a fuzzy linear assignment problem. It assigns each travel request to an available service vehicle through solving a linear assignment matrix of defuzzied cost entries. Each cost entry indicates the cost value of a travel request that "fuzzily aggregates" multiple criteria in simple rules incorporating human dispatching expertise. The approach is examined via extensive simulations anchored in a representative scenario of taxi deployment, and compared to the conventional case of using only distances (each from the taxi position to the source point and finally destination point of a travel request) as cost entries. Discussion in the context of related work examines the performance and practicality of the proposed approach.