Research On Automated Sanitation Vehicle Trajectory Planning And Tracking Control In Park Zone

An automated vehicle in the park zone is an integrated intelligent system that integrates many functions,such as environment perception,decision-making planning,and motion control.It is one of the first scenarios to be realized by automated driving due to its low speed,simple working conditions,and easy access to prior environmental information,which has high research value.Although the highly autonomous automated driving system in parks has been extensively studied and developed,many problems still need to be solved.Buildings and trees easily block positioning signals in the park zones.Therefore,more than traditional low-cost vehicle-mounted positioning equipment is needed to guarantee reliable accuracy and continuity of position data.As a result,the traditional low-cost vehicle positioning equipment usually cannot guarantee reliable accuracy or continuity of location data.At the same time,the mechanical structure and hardware deployment of automated sanitation vehicles in the park are subject to relevant restructuring and targeted design.The traditional vehicle dynamics model can produce deviation in describing the vehicle’s driving state.In addition,the sanitation vehicles in the park generally have practical problems such as slow steering and response.Passenger-carrying tasks in the park zone and trajectory tracking control links put higher steering smoothness and comfort requirements.The primary purpose of this paper is to solve the above problems in the driverless vehicle system in the park.The specific research contents are as follows:1)Fusion location algorithm based on multi-mode interactive filtering.The self-correcting lateral and longitudinal dynamic models that accurately describe the vehicle motion state are established.The dynamic compensation term is obtained by minimizing the estimation error of the modified model through the dynamic solution,which improves the estimation accuracy of the model and improves the strong dependence of the traditional prediction model on a single observation index.The movement mechanism of the restructured vehicles in the park is described from multiple perspectives through the vehicle motion model set.The information fusion architecture of multi-mode interactive filtering is built.The importance sampling scheme of expected positions and attitudes is designed to achieve full coverage predictive description of the accessible positions of vehicles.The vehicle motion model with centralized behavior mode is interactively estimated through particle filters.The reliability and continuity of the location estimation algorithm are verified by accurate vehicle testing.2)Multi-objective evaluation schemes of trajectory and dynamic environment modeling technology.The model describing the road section traffic cost under an a priori environment is established,and the bidirectional dynamic crossover search mechanism is constructed.The dynamic relaxation area determined by the proportion of search layers is designed as the criterion for interrupting the search process.The global trajectory planning test is completed under the actual high-definition map.The paper set up a reasonable frame of local trajectory based on terminal state sampling and the Bézier curve-based trajectory planning.The inversion probability model is established to reveal the relationship between the location estimation results and the environmental situation.A quantifiable virtual risk field and a probability grid model with a gradual resolution are designed to rapidly and accurately describe the local environmental situation and the global reference trajectory following the trend.Design trajectory multi-objective measurement strategy,establish the matching relationship between real-time motion trajectory and local environmental situation,and complete the multi-objective evaluation and optimization of the local reference trajectory.Finally,the planning performance of the local planner in the traffic scene is tested in a single frame to verify the feasibility of the description of the local environmental situation and the trajectory multi-objective evaluation scheme.3)The online self-correcting trajectory tracking control method imitates the driver’s natural driving behavior.A feedforward-feedback control scheme is established for trajectory tracking control.In the feedforward link,the fuzzy cerebellar neural network is introduced to modify the state evolution equation online so that the error-tracking model can gradually adapt to the actual motion state of the modified vehicles in the park.The feedback link constructs a game control strategy of "natural driving behavior error tracking," introduces constraints such as natural driving behavior characteristics and steering hysteresis of restructured vehicles into the control mechanism,and obtains the dynamic evolution information of the game environment.The controller can achieve smoother steering behavior with less tracking error by solving the equilibrium solution of the closed-loop Nash game.Carry out the joint simulation test of Carsim/Simulink to compare and analyze the controller’s performance to the steady-state error,the robustness of the road adhesion coefficient,and the vehicle speed under multiple scenarios and verify the feasibility.4)This paper is based on the automated driving system design scenario in the park to test.First,the overall scheme design with a particular type of engineering vehicle as the experimental platform has been carried out,and the relevant modifications and the construction of the vehicle platform have been completed.Then,according to the requirements of the real-vehicle test,the design of software architecture and real-vehicle test scheme is completed.The validity of the system in dynamic and static obstacle scenes,dangerous scenes during the steering process,and combination test scenes are verified by referring to relevant international standards.The experimental results show that the method described in this paper can assess driving risk in a dynamic and static environment and plan reasonable driving trajectories.It can effectively guarantee the accuracy and safety of the vehicle during autonomous driving.The validity of the method described in this paper is verified.