Improved SPEA2and Application On Robot Path Planning

Whether is in production or daily life, solving multiple objective optimization problem(MOP) is generally more difficult, the research on MOP has become the hotspot of operations research(OR) and the other related fields. Multi-objective evolutionary algorithm(MOEA) is a class bionic algorithm based on natural evolutionary process, and SPEA2is a classical algorithm which has been widely used in many fields. As a global algorithm, there is a lack of local search capability for SPEA2, and the diversity of evolutionary population will increase with the evolution.With the rapid development of technology, more and more mobile robots play an important role in production and life, and the robot navigation technology is one of the key topics in robot research. In fact, the robot path planning problem is a typical MOP which satisfy all the essential conditions of MOPs. Some main methods are unable to optimize multiple objectives simultaneously and exist their own defects, while SPEA2is very suitable for solving the robot path planning problems, SPEA2has a broad prospect in the path planning field.In this paper, the research work include the following aspects:(1) Most Multi-objective evolutionary algorithms focus on the global search capability too much, while generally pay less attention to the local search ability. In the paper, an improved SPEA2algorithm is proposed to make up for the weak point. An special external set is set used for local searching operation to ensure the strong local search capability. For further enhancing the algorithm convergence and maintaining the species diversity, the crossover operator is improved and the part individuals replacement strategy is joined in the SPEA2.(2)Classic SPEA2fixed Genetic parameters values, which limited the convergence rate, and even lead to the evolution of stagnation. In the paper An improved adaptive SPEA2algorithm is proposed, which involves an adaptive adjustment strategy which the genetic parameters values changed based on diversity of population to improve the convergence performance of the algorithm.(3)For applying the improved SPEA2based on the local search to the mobile robot path planning. Aiming at the path planning problem under complex environment existing multiple dynamic obstacles, the work environment model is established and the algorithm is designed in the paper. Under the MATLAB simulation platform, using traditional mixed target and the improved algorithm for the simulation of robot path planning.(4)Finally, the major contents of this paper is summarized, the existed deficiencies is analyzed, some issues still to be resolved is raised.