Research On Real-time Planning And Control Of Intelligent Vehicle Under Dilemma

With the development of technology and society,more and more intelligent vehicles are on the road at this stage.Although intelligent vehicles have a significant effect on the improvement of transportation pressure,safety performance and transportation efficiency.However,due to technical constraints and environmental uncertainty,the intelligent vehicles ensuring the safety of all pedestrians is difficult to achieve,and inevitably falls into a dilemma,it is very important for intelligent vehicles to avoid risks in a reasonable and real-time manner.The risk avoidance of the dilemma involving human harm will arouse public ethical disputes based on conventional ethical principles,and the quantification of the nominal cost model of conflict injury in line with public ethical values is the theoretical premise of reasonable risk avoidance.Under the dilemma,the planning and control module of the intelligent driving vehicle can achieve real-time,accurate and stable solution under the condition of the lowest nominal cost and the constraints of vehicle dynamics and vehicle stability,which plays a key role in solving the practical dilemma and avoidance problems.This paper mainly studies the real-time trajectory planning and tracking control module of intelligent vehicles under dilemma.The main research contents are as follows:(1)Construction of the nominal cost model.Based on the research group’s previous ethical research on the dilemma,this paper discusses the meaning and sensitivity factors of the nominal cost of conflict injury under the dilemma,and considering factors such as the degree of collision injury,regulations,ethics,and conflict uncertainty,a model of the nominal cost of human injury to all parties to the conflict is constructed.(2)Establishment and solution of the real-time trajectory planning problem in dilemma.The trajectory planning module in dilemma needs to minimize the nominal cost expectation while satisfying the constraints of vehicle physics and road boundary conditions.Since the nominal cost is expected to be a highly nonlinear and non-convex model,it is difficult to solve effectively and real-timely using the optimization algorithm in continuous space.In this paper,a trajectory planning algorithm based on state space sampling in discrete space is used to solve the dilemma planning problem effectively and in real-time.(3)Establishment and solution of real-time trajectory tracking problem.The dilemma is an emergency scenario.The tracking module in this scenario needs to consider the vehicle dynamics model with higher accuracy and the limit of vehicle slip stability.In order to solve the tracking problem involving complex models and multiple constraints,using Model Predictive Control algorithm(MPC)for a rolling solution.Considering the influence of nonlinear vehicle dynamics and slip stability on the realtime solution,online rolling linearization is used for processing,and the Linear TimeVarying Model Predictive Control(LTV MPC)is obtained,and then use Quadratic Programming(QP)to solve in real-time.(4)Real-time simulation experiment verification.Based on CARLA-ROS-Car Sim,a hardware-in-the-loop real-time co-simulation platform is built,and the trajectory planning and tracking control modules are experimentally verified.The results show that the trajectory planning module can get trajectory of the lowest nominal cost under the premise of ensuring robustness and real-time.The tracking control module has good real-time performance and robustness while ensuring small tracking error and satisfying vehicle stability.