Research On AGV Simultaneous Localization And Mapping Based On LiDAR

With the development of robotic technology,more and more robots have begun to step into industrial and household environments,and automatic guided vehicles(AGV),as an important part of industrial robots,have gained extensive application in recent years.The industrial application scenario environment is complex,and raises high requirements for the flexibility of AGV,therefore the simultaneous localization and mapping(SLAM)technology based on Li DAR was applied to this area.Under the influence of problems such as high motion noise,inaccurate modeling and particle dissipation,the predicted poses of the particles achieved by particle filter is poor.Based on the fact that each particle must maintain a global map separately,it is less effective to raise the number of the particles when the computing power of the vehicle hardware is limited.Therefore,it is of great theoretical and practical value to study how to improve the quality of the estimated poses of the particles,and reduce the influence of particle dissipation when the number of particles is limited.For the problem of Li DAR SLAM,this paper first introduces the internal and external sensors used in this project,analyzes two different odometer motion models,and adopts a differential-based feature extraction method,which is effective to extract point features from point clouds.Meanwhile,the Li DAR observation model was derived.Secondly,odometry and Li DAR data was used to solve the positioning problem by Kalman filter and particle filter.When solving SLAM problem by extended Kalman filter,a process noise variance determination strategy considering the rotation factor is adopted.Then,for particle filter based SLAM methods,a new proposal distribution was used to improve the poses of particles,as we ll as a new resampling strategy.Finally,in terms of the scan matching process in SLAM,this paper uses a correlative san matching method to improve the performance of particle filter based SLAM methods.The improved method is robust and not susceptible t o error accumulation.In order to verify the effectiveness of the improved strategy adopted in this study,both simulation and real experiments were designed.It was proved that the scan match method used in this subject effectively improved the pose quali ty of the particle,and its performance is better than other methods both in positioning accuracy and mapping result.Meanwhile,experiment under the robot operating system verifies the effectiveness of the improved proposed distribution.