Constraint-Based Scheduling
Generating Feasible Schedules for a Pick-Up and Delivery Problem
CP '98 Proceedings of the 4th International Conference on Principles and Practice of Constraint Programming
Ant colony system: a cooperative learning approach to the traveling salesman problem
IEEE Transactions on Evolutionary Computation
| Hi-index | 0.00 |
Transport problems are generally rather complex. The number of temporal, material, social and economic constraints makes it difficult to solved it both in theory (the problem is NP-complet) and in practice. This paper presents a pick-up and delivery transportation problem with time window and heterogeneous fleet of vehicles. Its objective is to provide an efficient schedule for drivers and the best service for users with the lower cost. This paper will explain the methodology used to compute and optimise the driver schedule. The goal is to assign more than eight hundreds fares to about forty drivers using thirty vehicles. Constraints programming approach makes it possible to treat instantaneously the local insertion of one journey in an optimised way. This local insertion was simply solved by an engine of constraints resolution in 1996 [MS88]. In addition, this method allows also the local insertion of several transports which was used to implement a total optimization of the schedule as a multitude of local displacements of few transports. This method becomes too slow when it moves more and more transport to improve quality of the solution. In this paper, a parallel solution, based on PVM, is introduced and some interesting results are provided. It is now possible to manage the collaboration of independent optimization engines working with different parameters. Experimentally, this robust solution is most of the time able to provide the best known sequential result.