Study On Autonomous Mobile Robot Trajectory Planning Based On Modified Quantum-behaved Particle Swarm

As an important component of autonomous navigation, Swarm intelligent algorithm based trajectory planning for Autonomous Mobile Robot (AMR) is getting more and more attention and has become a hot spot of research in recent years.In this paper, state of the swarm intelligent algorithm based AMR path planning has been surveyed, then AMR trajectory planning method based on Quantum-behaved Particle Swarm Optimization (QPSO) and its modified algorithms are proposed.The main contents of the paper include:(1) the main parameters relationship and convergence analysis of AMR path planning based on particle swarm optimization algorithm;(2) the convergence and control parameters research of quantum-behaved particle swarm optimization algorithm;(3) AMR trajectory planning based on quantum-behaved particle swarm optimization;(4) AMR path planning based on an improved natural selection quantum-behaved particle swarm optimization (SelQPSO);(5) AMR path planning based on a hybrid modified quantum-behaved particle swarm optimization (LTQPSO);(6) Trajectory planning for AMR navigation with a modified method in random obstacle environments.First, the evolution equation for standard PSO algorithm was changed a second-order non-homogeneous differential equations, the convergence property of PSO algorithm were studied in the swarm group space. Using the theory of probability theorem, the relationship between basic parameters convergence of PSO and the initial population distribution were analyzed. A new method of path planning was established through combine polar coordinates and rectangular coordinates, the feasible convergence region of the several improved PSO algorithm for the average distribution and normal distribution were calculated, the path planning for AMR based on several improved PSO algorithm were compared under different initial distributions. Simulation results have shown that a new path planning method is better than the classic method.Second, the convergence analysis and its modified method of QPSO algorithm was concentrated in the thesis. The parameters control strategy for QPSO was given, which provided the favorable control parameter strategy of QPSO in the simulation experiments. At the same time, the path planning method for AMR based on QPSO algorithm was proposed, and the relationship between the basic parameters and the average distribution with normal distribution was analyzed in resulting convergence region. A new method of trajectory planning for AMR based on QPSO algorithm was presented by introducing kinematics and dynamics. The performance of this new method was compared against trajectory tracking control based on path planning through using the autonomous mobile robot platform, experiment results demonstrated that the this method of trajectory planning for AMR based on QPSO algorithm has effectively achieved better tracking.Third, in order to enhance the global search ability and robustness of QPSO algorithm, a natural selection quantum-behaved particle swarm optimization (SelQPSO) is proposed. Several well-kwon benchmark functions are selected as testing functions. The two improved SelQPSO by difference natural selection numbers are tested and compared with the QPSO algorithm on the selected unimodal and mutimodal benchmark functions. Simulation experimental results have demonstrated preferable selection number and good performance of the SelQPSO algorithm. In addition, an improved SelQPSO algorithm is used in AMR path planning problem. According to initial distribution of QPSO and SelQPSO algorithm, the main parameters relationship are analyzed and formulated. The path planning for AMR based on SelQPSO algorithm with basic parameters estimation method is compared with QPSO algorithm for convergence speed and path quality. SelQPSO algorithm with basic parameters estimation method shows better performance and reliability than QPSO algorithm in AMR path planning application.Fourth, a hybrid improved QPSO (LTQPSO) algorithm is presented by combining QPSO with the individual particle evolutionary rate, swarm dispersion and natural selection method. In LTQPSO algorithm, the individual particle evolutionary rate and swarm dispersion are used to approximate the objective function around a current position with high quality in the search space. Natural selection method is used to update from the worst position to best position in the swarm. Experimental results on several well-known benchmark functions demonstrate that the proposed LTQPSO performs much better than QPSO and other variants of QPSO in terms of their convergence and stability. After that, a new trajectory planning method is designed based on LTQPSO to evaluate the effectiveness and feasibility of it on trajectory planning for AMR. The relationships between main parameters are analyzed according to initial distribution of LTQPSO and obtained a unary linear regression equation. It is compared with QPSO, WQPSO, and IQPSOS in the aspects of solution quality, robustness, and convergence property through numerical experiments. Experimental results have indicated that proposed algorithm is very effective than any other algorithm.Finally, for online AMR trajectory planning problem in the environment with random obstacles, based on the modified QPSO algorithms with basic parameter estimation equations of theoretical research, the real-time AMR trajectory tracking control system based on the results of this paper was developed used in AMR platform. The simulation results have shown that the proposed method are effective and can be used in the real-time path planning of mobile robots. Above all, it is indicated the research of this paper provided with practicality and availability in the environment with random obstacles.