Trajectory Planning Algorithm Based On Complex Optimization And Automatic Driving Platform

Automated driving is the unmanned driving technology for cars through computers and their surrounding systems.Perception,decision planning,and control are the three cores of autonomous vehicles.In this paper,a trajectory planning algorithm system based on complex optimization is designed by studying the trajectory planning module in automatic driving.First of all,for urban roads,open areas,trajectories and other scenes,the corresponding part of the path reference line planning is designed to be a non-convex optimization problem using graph search.In order to improve the conversion speed,it is feasible to use the macro method to adjust the optimization problem.Domain,convert it into a convex optimization problem,then decompose the trajectory planning solution into path planning and speed planning by means of the alternating direction multiplier method,and finally convert the convex optimization problem into the optimal function by appropriately designing the objective function and constraints of the optimization problem Form of secondary planning.The objective function considers the path smoothness and speed smoothness of the generated trajectory,and the constraints consider the collision constraints of obstacles and road boundaries,vehicle dynamics constraints,etc.Finally,the effective set method is alternately used in speed planning and path planning,and the final trajectory planning output is obtained.In order to verify related algorithms and support subsequent development,this paper improves and expands the software and hardware system of the autonomous driving platform on the basis of the original.Use the docke container management of the entire project's operating environment,set up the server as a public warehouse for code and mirroring,and synchronize code and mirroring of different devices.A series of automated scripts have been written to achieve rapid deployment and simultaneous update of the system.At the same time,this article expands the Gazebo simulation platform so that it can load different simulation environments and simulated vehicles according to different needs,and uses NVIDIA Jetson TX2 as the operating platform to build a 1/10 scale physical vehicle.Through simulation and real vehicle verification,the feasibility of the trajectory planning algorithm based on complex optimization described in this paper in practical application is proved,and the reliability of the autonomous driving platform described in this paper is also proved.