Research On Vertical And Diagonal Parking Trajectory Planning Algorithm

The rapid popularization of vehicles has brought convenience to transportation,but the car traffic behind the high growth is also faced with a series of problems,among which slow and difficult parking is typical.Due to the shortage of land resources,the parking space of urban parking lots becomes more narrow,and most of the planned parking spaces are vertical and slanted.Therefore,the current automatic parking system,which is a research hotspot,also takes vertical and slanted parking spaces as the test scene.As the brain of the automatic parking system,the planning module has a great impact on the parking performance index of the autonomous vehicle.Aiming at the above problems,this paper makes an in-depth research on the parking trajectory planning algorithm of vertical and diagonal parking spaces.The trajectory planning of vertical and diagonal parking spaces is decoupled from path planning and velocity planning.In order to meet the requirements of different parking scenarios and different starting positions in the process of automatic parking,a parking algorithm was proposed,which firstly solved the single-step parking into the feasible domain and then made path planning according to different scenarios.On this basis,the length of each planned path and the state of the starting position and the state of the ending position are solved.As the input of speed planning,quintic polynomial is used for speed planning.Firstly,taking vertical and diagonal parking spaces as scenes,the initial position and pose 3D regions that can generate one-step reverse parking into the garage under multiple constraints such as collision and steering ability were respectively solved,laying the foundation for parking path planning under different scenarios in the future.Then,the integrated path-velocity planning and smoothing based on quintic polynomial under multiple constraints are introduced in vertical and diagonal parking scenarios.The preliminary planning path was designed based on the feasible domain of single step parking solved in Chapter 2,and then smoothed the path based on the quintic polynomial.Then,the vertical parking speed planning was carried out based on the quintic polynomial,and the whole parking trajectory planning process was completed,and the preliminary simulation was verified by Matlab.Then,the automatic obstacle avoidance path planning of static obstacle parking scene is studied for vertical and diagonal parking spaces.Considering the collision constraints and vehicle steering ability constraints,the planned path of circular arc plus straight line is obtained.The planned kneading times automatically change with the change of vehicles,obstacles,roads and parking space parameters,and the preliminary simulation is verified by Matlab.Finally,due to the limited scene and research level,only real vehicle verification is conducted on trajectory planning under conventional vertical parking scenarios based on Apollo platform.Firstly,the test platform and the real vehicle transverse and longitudinal control algorithm are briefly introduced.Then,9 groups of different starting positions of vehicles are verified,and the real vehicle trajectory and the planned trajectory are compared and analyzed.The vertical parking path planning method proposed in this paper can plan a collision-free path satisfying the vehicle kinematic constraints at different starting positions of vehicles,and the kneading frequency is less.With short parking time consumption and small error between the position and the planned position at the end of parking,it has a good adaptability to the scene,and has certain reference value for the actual application of automatic parking.