Study On Localization And Mapping Of Laser SLAM For Electric Autonomous Race Car

With the increasing popularity of driverless cars,as one of the key technologies to achieve driverless cars,positioning and mapping technologies have also received more and more attention.SLAM(Simultaneous localization and mapping),is one of the key technology for autonomous navigation and environmental exploration of driverless vehicles.It means building a map while positioning itself in an unknown environment.The project relies on the National Natural Science Foundation of China(51675257),the Liaoning Provincial Key Research and Development Plan Guidance Program(2017106020),and the Liaoning Provincial High School Overseas Education Program(2018LNGXGJWPY-YB014),relying on the China University of Automotive Engineering,AIWAYS Automobile title to support the Formula Students Autonomous China,based on the Liaoning University of Technology driverless racing vehicle,the research on Positioning and Mapping of autonomous racing vehicle based on laser SLAM.The main researches are as follows.(1)Laser SLAM Definition Processing:Based on FSAC racing vehicle,define coordinate systems of vehicle,lidar and odometer,establish relationships between coordinate systems.Define the odometer model and calibrate the odometer using the least squares method.The lidar model is defined,and the lidar information is preprocessed for the working environment of the lidar to achieve simple filtering of obstacles outside the pile barrel.(2)Laser SLAM Front-EndTaking ICP and CSM algorithms as examples,the front-end registration algorithm of laser SLAM is deeply studied.Having an in-depth understanding of the nature of the front-end algorithm in a mathematically described way.And have a real-time data collection with FSAC racing cars,and the driverless racing vehicle was used to verify the ICP and CSM algorithms by the collected data.(3)Laser SLAM Back-EndFor the two kinds of back-end optimization frameworks: filter optimization method and based-on graph optimization method,the algorithms with high usage rate are selected for further study.In the filter optimization research,the introduction of Bayesian filtering,taking particle filtering as an example,combined with the laser SLAM itself,through mathematical description and reasoning,understands the filter back-end optimization from the essence of the filter.And using mathematical descriptions to explain the entire process of the based-on graph optimization method.(4)Map Creation By Laser SLAMPerform simulation simulation experiments and real vehicle construction experiments,build a “8-word surround” environment model and build a vehicle model for FSAC racing cars by ROS(robot operating system),and verify Gmapping and Cartographer mappingalgorithms.Real vehicle construction experiment,Based on the "8-word surround" site data collected by the real vehicle hardware platform,the ROS software platform is used to verify the algorithm verification experiment of the real vehicle.The research results show that based on FSAC racing cars,the laser SLAM method can be used to achieve accurate positioning,complete reasonable mapping,and provide a good foundation for ensuring that FSAC racing cars can achieve path planning.