Research On SLAM Mobile Robot's Path Planning And Trajectory Optimization

Path planning technology is a research hotspot of mobile robot technology.Aiming at the actual situation that the intelligent robot will have multiple target points in the global path planning,there will be the problem of the balance optimization of the robot path.The improved particle swarm algorithm based on reverse learning and the new objective function proposed in this thesis are used to make the robot obtain a safer and smoother path in the case of relatively optimal path;in order to avoid obstacles in local path planning,secondary obstacle avoidance and easy to fall into local optimal problems,this thesis uses improved Q learning basic algorithm,combined with dynamic object trajectory prediction method to achieve real-time dynamic obstacle avoidance of mobile robots.1.This thesis proposes an improved particle swarm optimization algorithm and the objective function of path planning.First,the algorithm uses a reverse learning strategy to initialize the particle population,and the corresponding formula is used to continuously iterate the algorithm while dynamically adjusting the two according to the actual situation of the iteration period.An important PSO parameter: inertial weight and learning factor.The addition of the reverse strategy effectively improves the search ability of the algorithm,because the initial population is the excellent particles after the screening,then it can also promote the convergence accuracy of the algorithm and improve the stability of the algorithm.In this thesis,an improved particle swarm optimization algorithm(OLIPSO)is used to solve the problem of optimal solution of the function of mobile robot path planning under multiple target points.At the end of this section,comparing the DAPSO,MOPSO and OLIPSO algorithms through simulation experiments,it can be concluded that the improved algorithm has better optimization capabilities,the robot trajectory also has better safety and smoothness,and improves the robot's multi-object point path planning efficiency.2.Compared with the traditional local path planning using DWA algorithm,this thesis introduces the idea of reinforcement learning,and proposes to solve an improved algorithm of autonomous planning based on Q learning algorithm to improve the obstacle avoidance ability of mobile robots.By designing a new Q-value table and reward and punishment mechanism,the corresponding actions are selected according todifferent area divisions and dynamic object action prediction.In the fourth chapter of this paper,the effectiveness of the improved Q algorithm is verified through simulation experiments.3.Completed a set of completely independent research and development of mobile robot experimental prototype from hardware to software,and carried out experimental verification.On the experimental operating platform based on the autonomously built robot operating system,experiments such as map construction,autonomous navigation,and path planning are completed,and the feasibility of path planning in the actual environment of mobile robots is analyzed and verified.Through the work in this thesis,the trajectory selection of mobile robots in path planning tasks is more reasonable and more efficient.The combination of reinforcement learning and pedestrian prediction has also been verified in simulations.