Research On Driving Assistance And Planning-control Strategy Based On Multi-mode For Bus

Since buses have the characteristics of long length,high center of mass,a massive load of variability,instability risk,they are more easily prone to traffic accidents compared to passenger cars.Studies indicate that more than 90% of accidents are due to driver's factors,optimizing the driver's manipulation through the development of driver assistance systems or intelligent driving techniques to replace drivers' operation can increase both buses driving safety,reduce traffic accident rate caused by driver's misoperation or unconsciousness.The advanced driver assistance systems currently commercialized on the market lack diversified designs for driver control characteristics and cannot adapt well to drivers with different driving behaviors.Trajectory planning technology is one of the core modules of intelligent driving.At present,the trajectory generated by the search-based trajectory planning algorithm is singular,and it is challenging to generate diverse trajectories based on time-varying traffic flow scenarios and personalized driving needs.This article aims to reduce the traffic accident rate caused by driver factors and meet individual driving.Starting from two intelligent technologies,assisting the driver and using intelligent driving technology to replace the driver,based on the multi-mode design idea,the multi-mode human-like driving assistance strategy considering the driving characteristics of bus lane departure and the multi-mode trajectory planning strategy in dynamic scenes are developed separately.The main research contents are as follows:(1)An Active Disturbance Rejection Control(ADRC)algorithm based on the optimization of the dual regulation strategy is developed,aiming at real-time and robustness of the control algorithm's problem applied in the intelligent driving strategy.To achieve the ADRC parameter tuning,the dual regulation strategy is designed by combining coarse adjustment and fine adjustment.In coarse adjustment section,the parameter adjustment principle is designed,determining each value ranges of preferably parameters,the influence law on the ADRC algorithm of various control parameters are explored by tuning parameters,obtaining the range of the preferably parameters and setting them as the constraints of subsequent genetic optimization algorithm.Based on the coarse adjustment,the genetic algorithm is carried out to achieve fine adjustment.In order to improve the optimization ability of genetic algorithm,use floating-point number encoding to optimize parameters,design fitness function that takes overshoot and response time into account,use analog cross binary operator to realize cross operation,and learning from the exploration and utilization ideas of reinforcement learning,use adaptive mutation operator to ensure population diversity and algorithm convergence.By changing the internal parameters of the controlled object and imposing different external disturbances amplitudes,the robust performance and anti-interference performance of the ADRC algorithm based on the dual adjustment strategy are verified.ADRC tracking performance and operating efficiency of the algorithm are also verified by tracking over different control algorithms paths at the same vehicle speed.Following the basis of above algorithm proposed,the stability control algorithm is developed based on the dynamic lateral load transfer ratio.The dynamic lateral load transfer rate is designed that changes with the steering wheel angle and vehicle speed,improving the bus instability warning ability,a three-degree-of-freedom bus model is built as the reference model,the ADRC algorithm developed by the dual adjustment strategy is used as the decision-making algorithm's to decide the compensating yaw moment,and the stability control of the bus is realized by means of differential braking,which provides the basis for the integration of the following the multi-mode human-like active steering control strategy.(2)Aiming at the research problem of diversified driving characteristics and single control of driving assistance strategy,the lane departure driving characteristics of buses are identified,and a multi-mode dynamic early warning algorithm and human-like active steering control strategy considering the lane departure driving characteristics of buses is established.In order to achieve the identification goal of lane departure driving characteristics of buses,the commercial vehicle intelligent driving platform built in this paper is used to collect lane departure driving data,analyze and extract characteristic parameters,and identify the lane departure driving characteristics of buses based on the obtained effective characteristic parameter sets.On this basis,the multi-mode time-domain early warning indicators for straight road and curved road are deduced respectively,and taking comprehensively identified bus lane departure driving characteristics,driver reaction time,impact on vehicles in adjacent lanes,and lane departure assist system related specifications and standards into account,the multi-mode spatial domain early warning indicators is designed.Based on multi-mode time domain early warning indicators and multi-mode spatial domain early warning indicators,using deviation speed as a switching condition,a multi-mode dynamic early warning algorithm that integrates time domain and space domain is designed.Based on the electronic-hydraulic power steering system of commercial vehicles built,a coordinated control strategy for the power steering mode and the active steering mode is designed.In the active steering mode,in order to realize the human-like idea of "car adapts to person",based on the identification of the driving characteristics of the lane departure of the bus,a multi-mode human-like active steering control strategy that matches the driving characteristics is established;a co-simulation software platform combining Truck Sim,AMESim and MATLAB/Simulink software is built,and the straight road and curved road working conditions are selected to verify the effectiveness of the proposed multi-mode dynamic warning algorithm and multi-mode human-like active steering control strategy.(3)This paper develops a multi-mode trajectory planning strategy based on swarm intelligence algorithm optimization aiming at the singular planning trajectory problem in intelligent driving.In the static scenarios,in response to the large randomness and low efficiency of fast search random tree sampling,a bidirectional region sampling random tree search algorithm is proposed,which uses gaussian sampling randomness and local sampling directionality to accelerate the search of effective nodes.In order to improve collision avoidance,the detection efficiency is based on the intersection detection idea of oriented bounding boxes in different directions in computer graphics,and the separation axis theorem is used to realize the collision avoidance detection considering the actual size and dynamic characteristics of the bus.In order to reduce the vibration of the planned path,based on the steering and braking characteristic,two-step smooth strategy is designed.The planned path is generated by the initial smoothing principle followed by eliminating null point and driving consensus.On the basis,using the cubic B spline realize the secondary path smoothing.In the dynamic scenarios,in order to reduce the difficulty of trajectory planning,the Frenet coordinate system is used to decouple the longitudinal and lateral coupled trajectory planning problem in the two-dimensional space into two one-dimensional trajectory planning problems in the longitudinal and lateral spaces,using the fifth degree polynomial and the quartic polynomials respectively fitting the longitudinal and lateral trajectories to generate candidate trajectories.In order to improve the search efficiency of the optimal trajectory,the longitudinal and lateral terminal behavior planning strategies are designed respectively.To evaluate the generated candidate trajectories,based on the five cost functions of smoothness performance,comfort performance,efficiency,distance deviation,and speed tracking performance are designed,the genetic algorithm is used to optimize the fitness functions targeting comfort,efficiency and economy,and multi-mode trajectory planning strategy is achieved.Integrating dual ADRC(DADRC)algorithm,driving strategy and braking strategy,the longitudinal tracking control strategy is developed,combined with the ADRC path tracking control algorithm based on the dual adjustment strategy achieve effective tracking of multi-mode trajectories.(4)In order to realize the identification of lane departure driving characteristics of buses and the verification of the strategy developed in this paper,a commercial vehicle intelligent driving platform is built and the effectiveness of the proposed strategy and algorithm are verified.According to the requirements of identification of lane departure driving characteristics of buses and verification of the strategy proposed in this paper,based on Truck Sim vehicle dynamics simulation software,MATLAB /Simulink modeling software,the actual vehicle hardwares such as electronic-hydraulic power steering system,electronically braking system,real-time simulation systems developed by NI and d SPACE,the commercial vehicle intelligent driving platforms is built.The selection of hardware equipment related components is completed,the working principles of the electronic-hydraulic power steering system,the electronically controlled power steering system,and the real-time simulation system,hardware connection and system communication are introduced in detail.The commercial vehicle intelligent driving platform established realizes the effective communication of multiple software and hardware devices,realizes the identification of lane departure driving characteristics of buses,and effectively verifies the stability control strategy and the multi-mode human-like active steering control strategy for bus.