Research On SLAM And Navigation Of Mobile Robot Based On Sensor Fusion

In recent years,with the rapid development of robots in our country,robots often appear in our daily life.Due to the complexity and uncertainty of the surrounding environment in the process of moving,mobile robots are challenged to be more intelligent in the future.Applied to the indoor environment,the three key technologies of mobile robots,namely,realtime positioning,map construction and path planning are studied.In view of the characteristics that mobile robots can move flexibly in complex indoor environments,this paper builds and optimizes environmental maps based on the robot operating system platform.The autonomous navigation function of the mobile robot is realized by using the improved positioning algorithm and path planning algorithm.The main research contents of this paper are as follows:(1)Analyze and optimize real-time positioning and map building algorithms.Firstly,the traditional RBPF-SLAM algorithm is studied.As the motion model is needed in particle filter algorithm,and the motion model provided by the odometer has a large error.The unscented Kalman filter algorithm is used to fuse the information of the inertial measurement unit and wheel odometer information as the motion model in the system.Secondly,according to the particle degradation phenomenon in the particle filter algorithm,the motion model and the lidar observation model are combined in RBPF-SLAM to improve the proposed distribution,and the annealing parameter is added to prevent the peak distribution in the observation model,which will affect the sampling efficiency.Finally,in order to improve the diversity of new particles,the Gaussian distribution is used for resampling,and the particles are sorted according to the weights and then dispersed.To verify the performance of the algorithm,two SLAM data sets are run under the ROS platform to verify the effectiveness of the algorithm.(2)Design improvement plans according to the problems existing in path planning.First,in the research of global path planning of mobile robots,aiming at the problem of discontinuous path curvature generated by the A* algorithm,a new heuristic function is designed by combining Manhattan distance and Euclidean distance.And gradient descent method is used to smooth the global paths which with too many inflection points.Second,the temporal elastic band algorithm is used in the local path planning,which enables the mobile robot to cope with unexpected situations in practical applications.Finally,by fusing global path planning and local path planning,the static and dynamic obstacle avoidance is realized.In the simulation environment,the feasibility of the path planning algorithm described in this paper is verified.(3)For the indoor scene of robot application,experiments on map building and path planning are carried out on the mobile robot platform in two environments in the laboratory,and unknown obstacles are added to test the obstacle avoidance performance of the mobile robot.Repeated experiments are carried out to verify the effectiveness of the improved RBPF-SLAM algorithm and the feasibility of the overall navigation function of the mobile robot.