Tabu Search for Driver Scheduling

Abstract

This paper presents a Tabu Search heuristic for driver scheduling problems, which are known to be NP-hard. Multi-neighbourhoods and an appropriate memory scheme, which are essential elements of Tabu Search, have been designed and tailored for the driver scheduling problem. Alternative designs have been tested and compared with best known solutions drawn from real-life data sets. The algorithm is very fast, has achieved results comparable to those based on mathematical programming approaches, and has many potentials for future developments.