Research On Ant Colony Optimization Algorithms For Path Planning Of Mobile Robot

Path planning of mobile robot is one of important research area of robot techniques, which has the characteristics of complexity, binding nature, nonlinearity, so the research of path planning algorithms is in the ascendant recent years. Existing optimization algorithms have many defects in solving path planning problem under complex environment, such as poor robustness, non-optimal and low efficiency, but ant colony optimization algorithm (ACO) is a new intelligent optimization algorithm with the advantages of strong robustness, implicit parallelism, global optimization performance, which can combine with other intelligence methods easily. And its biological mechanism is ant colony searches a shortest and feasible path between ant nests to food source, which coincides with physical process of path planning for mobile robot and provides important basis for the rationality of path planning research based on ant colony optimization algorithm.This dissertation focuses on the improvements of basic ACO algorithm and the application of basic ACO algorithm on path planning of mobile robot, based on the principles and theory of ACO algorithm. The main achievements of it include:(1) Firstly two-dimensional environment are modeled by grids method combining graph theory in this paper, and then the models of three-dimensional space are obtained by utilizing space tangent plane sharing method.(2) It is formulated on the choice of parameters of basis ACO algorithm by theoretical analysis and simulation experiments, and then the influence of algorithm parameters on algorithm performance is summarized and the suitable ranges of algorithm parameters are given, which laid the groundwork for later algorithm studies.(3) ACO algorithm with fuzzy adaptive window characteristic (FWACO) is proposed since basis ACO algorithm easily gets into local optima, searches slowly, and lacks precise theory direction for parameters selection. This method designs fuzzy controllers to optimize three important parameters of α、β and(1-ρ), and the initial pheromone is distributed reasonably to increase efficiency. In addition, the concept of active degree of node is presented as future information to direct ants selecting path, and a new heuristic method for pheromone updating is introduced. Later, the new path evaluation index is given to overcome the shortage of losing problem factuality if only the path length is used as evaluation function. Finally many simulation researches are made under two-dimensional environment and three-dimensional space, and simulation results indicate this method has fast convergence speed and can plan optimal path or approximate optimal path even in complex environment.(4) Differential evolution chaos ant colony optimization (DECACO) algorithm is introduced to plan the optimal collision-free path for a mobile robot in complicated environments. It utilizes differential evolution algorithm to update the pheromone, appends the chaos disturbance in updating process to avoid the possible stagnation stage and then a new evaluation criterion is employed, which enhance the escaping capability of algorithm, refrain from the path-locked situation and improve the searching efficiency for optimal path. Finally many simulation researches are made under two-dimensional environment and three-dimensional space, simulation results show this method has strong robustness, better adaptability for environment barriers, and an optimal and safe path on which the robot moves on can be obtained even in a complex environment.(5) Two-way parallel search ACO (TWPSACO) algorithm proposed by others is improved based on searches in last two chapters, and is used in the path planning problem of mobile robot. Since it is clearly seen that TWPSACO algorithm has a fatal defect of losing some feasible paths and even losing optimal path, so a new ants meeting judgment method is proposed. Then a new path selecting method and a new global pheromone updating technique are proposed in order to avoid running into local optima. At the same time, two-dimensional environment models of mobile robot are established by grids method with coding method based on effective vertexes of barriers in order to avoid environment traps. Simulation researches show as long as the path objectively exists the algorithm can plan a safe optimal path quickly, and the effect is satisfactory.Overall, these three improved ACO algorithm given in this dissertation is valid and feasible for path planning problem of mobile robot, and they have accurate optimization performance, quicker searching speed, strong robustness and better adaptability on environment barriers, so their comprehensive properties are desirable. Finally, the works of this dissertation are summarized, and then some problems, which need to be further studied, are brought forward.