Research On Local Path Planning Of Intelligent Vehicle Based On Multidimensional Lidar

Under the background of the rapid development of computer and artificial intelligence technology,the research and development of intelligent vehicle has received extensive attention.The major Internet companies and automobile enterprises have invested a lot of energy in the research and development of intelligent vehicle technology.As one of the new sensors,lidar plays an important role in the development of autopilot technology and becomes the most reliable sensor for autopilot map construction and obstacle recognition.As the input of the whole intelligent driving system,the perception module has a great influence on the subsequent decision-making,planning and control.At the same time,in order to ensure the safe driving of vehicles,local path planning has become an indispensable link.Therefore,it is of great significance to study the separation method of laser point clouds and local path planning for the safe driving of intelligent vehicles.In order to avoid obstacles smoothly,the size and relative speed of obstacles must be accurately identified first.The points hit by laser on the ground will be mixed with obstacles,which makes it impossible to identify obstacles accurately.Therefore,the point clouds of the ground and obstacles should be separated to improve the accuracy of obstacle recognition.In this paper,a new algorithm for separating point clouds on the ground is proposed on the basis of three-dimensional laser point cloud grids.In this method,the three indexes of radial distance,elevation difference and gradient are selected,and the attributes of laser points are judged by setting a threshold,so that the ground points can be found and finally the ground points can be separated.Local path planning algorithm is based on vehicle dynamics and related theory,such as traditional robot trajectory planning will ackerman steering type of vehicle as the research object,through to the rapidly expanding random tree(RRT/rapidly exploring random tree)algorithm was improved,the nonholonomic constraint to join the sampling algorithm,make planning of the obstacle avoidance path to meet the requirements of vehicle steering constraint,and at the same time by curve fitting method for the path ofquadratic optimization,to ensure that the planning of the trajectory is more smooth and improve the stability of vehicle.Finally,the experimental car platform is designed to verify the actual effect of the above algorithm.Through the analysis of the experimental results in different scenarios,it is found that the algorithm has good performance.The ground point cloud separation algorithm can effectively separate the ground and obstacle point clouds in different scenarios on the premise that the information of obstacles is not distorted;the local path planning algorithm can generate smoother obstacle avoidance paths.Finally,compared with other algorithms in terms of efficiency,scene versatility and so on,it proves that the proposed algorithm has better effect.