Research On Mobile Robot Path Planning Algorithm

As one of the key technologies for mobile robot systems to perform tasks,path planning algorithm has always been the focus of research in this field.The path planning algorithm of mobile robot is divided into global path planning in which all environmental information is known and local path planning in which part of the environmental information is known,respectively.This thesis studies the path planning algorithm of mobile robots in complex environments as follows:(1)This thesis establishes a mathematical model for the path planning problem of mobile robots.Based on the fixed length division method of the coordinate system in the environment,the decision variables and objective functions of the path planning problem are determined.The hardware platform and software platform to be studied are introduced,while the kinematics model is established for the specific mobile platform with analysis of the kinematic constraints.(2)Focusing on global path planning,an improved particle swarm optimization(PSO)algorithm is proposed.In order to solve the problem that PSO algorithm is easy to fall into local optimum,a mutation operator that varies with the number of iterations of PSO algorithm is designed.At the initial stage of the iteration,updating the position of particles by small-range mutation search can make the algorithm converge quickly.At the latter stage of the iteration,the particle position is updated in the global scope,such that the local optimum is prevented and the global optimum can be thereby achieved.(3)As for local path planning problem of mobile robots,feedback compensation neural network(FCNN)is proposed.The algorithm proposes a new neural network topology,which can be seen as an improvement of the traditional back propagation neural network(BPNN).The first BPNN is properly trained,playing a dominant role in dynamic obstacle avoidance.The second BPNN is trained online in order to obtain the compensation for the output during the movement of mobile robot.The FCNN can predict the motion of dynamic obstacles according to the previous obstacle avoidance situation and reduce the number of decision-making and the burden of operation in robotic systems,even if the obstacle velocity continuously changes.(4)For the improved PSO algorithm proposed in this thesis,the simulation of global path planning is carried out in the complex environment of multiple obstacles,and the performance is analyzed.According to the specific environment,determining the parameters of the optimal number of iterations,size of particle swarm and number of path nodes.As compared to the traditional PSO algorithm,the improved PSO algorithm has been greatly improved from both convergence speed and global optimization ability aspects.The simulation studies of the obstacle avoidance performance of the proposed local path planning algorithm FCNN in a dynamic environment are carried out.Compared with the traditional BPNN,the FCNN algorithm has obvious advantages in path length and number of decisions.Moreover,the effectiveness of the proposed improved PSO algorithm is validated via experiments based on a four-wheel mobile robot.