New heuristic and evolutionary operators for the multi-objective urban transit routing problem

The urban transit routing problem (UTRP) involves finding efficient routes in a public transport system. However, developing effective heuristics and metaheuristics for the UTRP is hugely challenging because of the vast search space and multiple constraints that make even the attainment of feasible results exceedingly difficult, as the problem size increases. Moreover, progress with academic research on the UTRP appears to be seriously hampered by: 1) a lack of benchmark data, and 2) the complex and diverse range of methods used in the literature to evaluate solution quality. It is not currently possible for researchers to effectively compare the performance of their algorithms with anyone else's. This paper presents new problem-specific genetic operators within a multi-objective evolutionary framework, and furthermore proposes an effective and efficient heuristic method for seeding the population with feasible route sets. In addition new data sets are provided and made available for download, to aid future researchers. Excellent results are presented for Mandl's problem, which is currently the only benchmark available, while the results obtained for the new data sets provide a challenge for future researchers to beat