Research On Smart Car SLAM Navigation Based On Multi-sensor Fusion

The rapid development of artificial intelligence,big data,and cloud computing has changed the form of the traditional automobile industry.Autonomous driving is a hot topic of current research.Its core technology SLAM(Simultaneous Localization and Mapping)and autonomous navigation have broad prospects and markets.Research on the application of related algorithm technology to smart cars has far-reaching guiding significance for the development of smart cars.The thesis takes the smart car equipped with lidar,odometer,and IMU sensors as the research object,and studies the problems of synchronous positioning mapping(SLAM)and autonomous navigation in unknown environments.First,systematically analyze the overall architecture of the smart car,and establish related mathematical models,including the speed movement model of the smart car,the odometer movement model,the IMU model and the lidar observation model;synchronous positioning and mapping,focusing on the Cartographer based on graph optimization-2d-SLAM algorithm,in order to solve the problem of inaccurate posture estimation of single odometer information in the original algorithm,IMU sensor information is added to the original laser odometer model,and the odometer and IMU information are integrated through the lossless Kalman filter(UKF)to make small parking spaces Attitude prediction,combined with laser scanning matching to achieve multi-sensor fusion mapping,and use MIT data set for simulation test,the test shows that the multi-sensor fusion mapping effect after adding IMU information is better than the original laser odometer model mapping effect;autonomous navigation In the process,Dijkstra algorithm and A* algorithm are used for global path planning,and simulation tests are carried out.The results show that A* algorithm uses heuristic function to search for target points,and its search efficiency is much higher than Dijkstra algorithm.It uses dynamic window method(DWA)The algorithm is used for local path planning.Aiming at the problem of dynamic window method detouring and deviating from the global path in local navigation,a trajectory evaluation function based on the global path is proposed,and the distance from the estimated trajectory to the nearest global path point is added to the original trajectory evaluation function Based on the ROS operating system,the smart car is equipped with lidar,odometer,and IMU sensors to establish an experimental platform,and simulation tests show that the optimized DWA algorithm can choose a simpler trajectory route.The IMU deterministic error calibration is performed on this platform,and the SLAM autonomous navigation experiment of the smart car is carried out in combination with the improved SLAM navigation algorithm.The experimental results show that the multi-sensorfusion mapping with IMU information added in the process of synchronous positioning and mapping has higher mapping accuracy and stronger robustness;in the process of autonomous navigation,the optimized DWA algorithm makes the smart car in When local obstacle avoidance can quickly converge to the global path,the path selection is better.