New Model for the Multiple Vehicle Path Planning Problem and GA-based Solutions

Recent years, research on intelligent transportation systems (ITS) attracts a lot of attention. One of key problems of ITS is the multiple vehicle path planning (MVFP) problem, which requires to find (nearly) optimal path schedules for multiple vehicles, might unmanned, over a directed graph such that all target points are covered by the union paths of these vehicles and meanwhile to ensure that each vehicles has visited its own target points. This problem is a generalization of the Hamiltonian path problem, with more complex constraints, and is obviously NP-hard. In this paper, we propose a novel genetic algorithm for solving the MVPP problem based on a Lagrangian relaxation technique. New genetic operators, including crossover, mutation, and adaptive penalty and mature-degree control strategies are proposed. Experiments show that our proposal is effective.