Investigation Of Control Strategies For Intelligent Vehicles In Dangerous Traffic Scenarios

Based on autonomous driving systems,intelligent vehicles can realize automatically driv-ing,which is useful in reducing traffic accidents,easing traffic jams,and reducing environmental pollution.At home and abroad,the investigation of intelligent vehicles has become a hot point both in industry and academia.Active safety technology is one of the key problems in the development and application of intelligent vehicles.Comparing with traditional human driv-ing vehicles,intelligent vehicles,which are equipped with autonomous driving systems,can completely control the vehicle motion.This also leads to a higher demand for the active safety technologies of intelligent vehicles.How to design complete control strategies for intelligent vehicles is the main problem in the development of active safety technologies.Currently,some dangerous traffic scenarios are not investigated completely.Some im-provements may be obtained based on intelligent vehicle technologies to improve traffic safety.In this study,an extension of traditional vehicle active safety technologies is made.The con-ceptions of active escape,active deceleration and active collision are proposed,which extend the research field of vehicle active safety technologies.Moreover,several problems relevant to the proposed conceptions are investigated.Some corresponding control strategies are made for these problems.The researches are as follows:(1)The conceptions of active escape,active deceleration and active collision are proposed.Active escape means that an intelligent vehicle should try to avoid the traffic accidents caused by other vehicles.Currently,the collision avoidance control systems of intelligent vehicles include adaptive cruise control,rear-end collision avoidance,emergency braking,etc.However,most of these systems focus on reducing the traffic accidents caused by host vehicles.How to avoid the traffic accidents caused by other vehicles is generally ignored. These systems may fail to handle some dangerous traffic scenarios,such as rear-ended by other heavy duty vehicles.To further improve the safety of intelligent vehicles,the conception of active escape is proposed.The longitudinal control method to avoid the collision accidents caused by other vehicles is studied.Two problems are considered:how to avoid the rear-end collisions caused by other vehicles and how to avoid the frontal collisions caused by retrograde vehicles.Active deceleration means that,when the risk of side-slip in the front road is detected,an intelligent vehicle should actively decelerate to reduce the risk of side-slip.Current investigations on yaw stability control are helpful in reducing side-slip accidents.However,most of these investigations are valid only with the occurrence or about to occurrence of side-slip.The pre-deceleration control is generally ignored.To resolve this problem,the conception of active deceleration is suggested,which focuses on reducing side-slip accidents by friction coefficient estimation and speed planning.Active collision means that an authorized intelligent vehicle should actively collide with an invaded vehicle to avoid severe social harmfulness.With the development of intelligent vehicles and vehicular network,the risk of network security is becoming serious.Most of the researches on network security focused on network communication.Few considered how to prevent the vehicle terrorist attacks caused by invaded intelligent vehicles.Moreover,the common police equipments are difficult to block such dangerous vehicles with social risks.For this problem,an active collision controller is proposed in this study,which is useful to control the host intelligent vehicle to crash the invaded dangerous vehicle by active collision.(2)A rear-end escape control system and a front collision escape control system are designed for the dangerous traffic scenarios with the risks of rear-end or frontal collisions.A longitudinal dynamics model of four wheels independent drive vehicle is built considering the air drag force and the rolling resistant force.Based on this model and model predictive control algorithm,a model predictive motion controller is constructed to control the longitudinal motion.To reduce the calculation burden,a longitudinal motion controller is also developed using the PID algorithm.According to the analysis of rear-end collisions on single lane roads,the rear-end collision escape control system is proposed based on the flocking algorithm of multi-agent systems.For the single vehicle rear-end escape problem,a single layer multi-agent flocking structure,which consists of ?-agent,?-agent and y-agent,is constructed.Based on the interactive relationship among agents,a control protocol is introduced considering speed tracking and rear-end avoidance.A trajectory planning algorithm for single vehicle rear-end escape is designed based on the flocking algorithm.For the multiple vehicles rear-end escape problem,a two layers multi-agent flocking structure is developed.Based on the interaction among the agents between two layers,a trajectory planning algorithm is investigated for the rear-end collision escape of vehicle platoon,which is useful to coordinate multiple vehicles to reduce rear-end collisions.Based on the analysis of frontal collision,the decision logic to judge the risk of frontal collision is provided.According to the analysis of collision risk,the trajectory planning algorithm to escape front collision is proposed for a single-vehicle.A multi-vehicle cooperative trajectory planning algorithm is investigated,which is useful to coordinate the next vehicle to reduce the risk of frontal collisions.To further improve the safety of vehicles,a multi-vehicle optimal cooperative trajectory planning algorithm is designed,which is effective to fully exploit the performances of different vehicles and further reduce the frontal collision accidents of heterogeneous vehicles.Simulations are applied to verify the effectiveness of the proposed control system for single or multiple vehicles in escaping rear-end and frontal collisions.(3)Based on friction coefficient estimation,a speed planning algorithm is proposed for the dangerous traffic scenarios with the risk of side-slip.On the road with a high curvature and a low friction coefficient,due to the limit of tire forces,high speeds may lead to traffic accidents such as side-slip.However,few investigations have been made based on active deceleration to reduce the risk of side-slip in such a scenario.To resolve this problem,a speed planning algorithm is investigated.A seven degree-of-freedoms longitudinal and lateral coupling vehicle model is constructed,which includes the longitudinal motion,lateral motion,yaw motion and wheel rolling motion.Based on the wheel dynamic model,chassis dynamic model and wheel deformable model,the estimation methods to calculate longitudinal and lateral tire forces,wheel effective radius,slip-ratio and side-slip angle are designed.Using the Pacejak tire model,the effect of the tire force utilization ratio to friction coefficient estimation is analyzed.The relationship between the friction coefficient and the tire forces on common motion scenarios is discussed.An iterative optimization algorithm for friction coefficient estimation is proposed.Considering the effect of tire slip ratio to friction coefficient estimation,an adaptive torque injection method is proposed to effectively estimate the friction coefficient on the scenarios without hard acceleration or steering.Based on the division of the frontal path with a constant distance,a speed optimization algorithm is proposed considering side-slip,roll-over and dynamic constraints.The quadratic programming problem for speed optimization is also provided.The proposed speed planning algorithm can estimate the friction coefficient and plan a safe speed,which is useful to slow down the vehicles on the road with a low friction coefficient.The proposed algorithm is helpful to reduce the side-slip accidents on the road with a low friction coefficient.The simulation to test the effectiveness of the proposed speed planning algorithm in reducing the risk of side-slip accidents is made.(4)A vehicle active collision controller is proposed for the dangerous traffic scenarios with the invaded vehicle and the risk of vehicle terrorist attacks.The development of intelligent and connected vehicles significantly facilitates the public.However,this also increases the risk of network intrusions.In American a vehicle was hacked and controlled,which leaded to the recall of related vehicles.This also increases the risk of severe vehicle terrorist attacks at home and abroad.While the common police equipments are difficult to cope with such terrorist attacks.To resolve this problem,in this study,an active collision controller is designed for intelligent vehicles.With authorization,this controller can be used to control intelligent vehicles to active collide with invaded and dangerous vehicles.Considering the longitudinal motion,lateral motion and yaw motion,a three degree-of-freedoms longitudinal and lateral coupling vehicle model is developed.Based on the analysis of vehicle relative motion,the equation for active collision is obtained.Using the first-order linearization in the operation point,a linear model for the controller design is constructed.Based on the framework of model predictive control,an active collision controller is designed.Using simulation,the designed controller is proved useful to collide with other vehicles with different motions.(5)A miniature intelligent model vehicle,which consists of steer-by-wire and drive\brake-by-wire,is built for the experiment.The chassis of the miniature intelligent model vehicle is constructed with industrial aluminum.The driving system is built using DC motors and a motor driver.The steering system is developed with steering motors and RS485 bus.A USB data acquisition board is used for signal acquisition and motor control.In a laptop,the decision making,trajectory planning and motion controller of the frontal collision escape control system are constructed using C++based on the framework of Microsoft foundation classes.An experiment scenario is conducted to show the effectiveness of the designed frontal collision escape control system.The main contribution of this study is the extension of the research field of vehicle active safety.The conceptions of active escape,active deceleration and active collision are proposed.The control strategies are designed for several dangerous traffic scenarios.The main innovations are followed:(1)The rear-end escape and frontal collision escape control systems for intelligent vehicles with the vehicular network are firstly studied.The proposed control systems can control single or multiple vehicles to avoid rear-end and frontal collision accidents by accelerating,braking and reversing.Most of the traditional investigations of vehicle collision avoidance focused on reducing the collision accidents caused by host vehicles.In this study,the rear-end and frontal collision accidents caused by other vehicles are further considered.(2)Combing with friction coefficient estimation,the speed planning algorithm is firstly proposed.This algorithm can plan a safe speed based on the on-line estimated friction coefficient,which is useful to slow down the speed with the detection of side-slip risk.Most of the relevant researches focused on reducing side-slip accidents by steering or yaw movement control.Few investigations considered speed planning based on the on-line estimated friction coefficient.(3)The active collision controller is firstly designed.The controller can be used to control intelligent vehicles to collide with social harmful vehicles.Few investigations have been made for the designing of an active collision controller.This paper is helpful for the promotion and application of intelligent vehicles,which is also useful to improve traffic safety.