Research On Lateral Control For Highway Navigation Pilot Oriented Towards Autonomous Driving Safety Requirements

Highway navigation pilot(HNP)is an automatic assisted navigation driving system that enables vehicles to navigate from point to point within a closed structured road range under the supervision of the driver according to the route planned for navigation,such as highways and urban expressways.HNP system is one of the mainstream autonomous driving systems at present,in which the lateral control plays a vital role in driving and is of great significance for achieving a safe,stable and efficient driving experience.Despite its noteworthy accomplishments in the realms of perception,planning,and control,the HNP system,which operates as a higher-order assisted driving system between level 2 and level 3,still confronts numerous aspects that need to be improved.With the development of intelligence and networking,the electronic and electrical systems of autonomous driving vehicles are becoming increasingly complex,so it is necessary to consider the functional safety issues of lateral control caused by electronic and electrical faults.Furthermore,the complexity and diversity of objectives,as well as the uncertainty of the social environment,may lead to potential system design deficiencies or sensor performance limitations in HNP,resulting in safety of the intended functionality(SOTIF)issues of lateral control.In addition,with the increase in the number of sensors in autonomous vehicles and the use of advanced in vehicle communication technologies,it is crucial to consider the lateral control safety issues that arise from cyber attacks.This paper takes autonomous vehicle as the research object.A safety analysis is conducted on the lateral control of HNP from the perspective of functional safety and SOTIF.Based on this analysis result,safety requirements for Electric Power Steering(EPS)are proposed.In response to safety requirements,an adaptive fault-tolerate lateral controller is designed accordingly.In addition,from the perspective of cyber security,an adaptive secure switching lateral controller under Denial-of-Service(Do S)attacks is designed.The main contents of this paper are as follows.(1)To address the issue that Fault Tree Analysis(FTA)inadequately derives safety requirements when conducting functional safety analysis of HNP lateral control,focusing solely on analysis for failure causes,a functional safety analysis method for HNP lateral control based on the Safety Goal Structured Analysis(SGSA)is proposed.Initially,the elements of the SGSA are described,including backgrounds,assumptions,and judgments.Subsequently,the safety strategies supporting safety goal for HNP lateral control are derived using SGSA.Finally,a swimlane diagram is employed to illustrate the allocation of safety strategies on architectural elements,and the safety requirements for the architectural elements are described.Among them,the safety strategy assigned to the EPS is that EPS should have sufficient stability and doesn’t occur unexpected steering in the event of internal faults.(2)To address the issue that Systems-Theoretic Process Analysis(STPA)fails to provide causal scenarios identification when conducting SOTIF analysis of HNP lateral control,an integrated approach combining STPA with Cause Tree Analysis(CTA)for HNP lateral control is proposed.Initially,the analysis process of STPA and CTA are introduced.Subsequently,STPA is utilized to identify unsafe control actions and evaluate SOTIF risks,while CTA is employed to analyze causal scenarios.Finally,corresponding SOTIF requirements are proposed for different components based on their limitations.Among them,the safety requirement for the EPS is that EPS should have sufficient robustness to avoid unexpected steering on slippery and winding roads,potholed roads or stone impacts.(3)To satisfy the safety requirements of EPS from the perspectives of functional safety and SOTIF,an adaptive fault-tolerant lateral controller for time-varying unknown parameters and intermittent actuator faults is proposed.Initially,the lateral tracking error model of autonomous vehicle is derived from the kinematics model and dynamics model of vehicle lateral motion.Subsequently,a new adaptive control scheme is designed by introducing an adaptive quadratic damping term to estimate and compensate for time-varying unknown parameters and fault coefficients.Finally,the effectiveness of the control scheme is verified through simulation experiments.Compared with PID controller,sliding mode controller and Backstepping BLF controller,the designed controller can guarantee system performance under intermittent actuator faults and sudden changes in unknown parameters.The designed control scheme is capable of achieving the desired control objectives.(4)To address the issue of lateral control instability caused by the inability of vehicle to obtain accurate position and pose information when the Global Navigation Satellite System(GNSS)equipped with the vehicle is subjected to Do S attacks,a controller with active defense capability is designed.Initially,considering that Do S attacks are implemented intermittently on GNSS sensors.Subsequently,an estimator is introduced and the position and pose information of vehicles are estimated based on the linear velocity and angular velocity data provided by the Inertial Measurement Unit(IMU).A novel adaptive secure switching control scheme is designed using the backstepping method.Finally,the effectiveness of the control scheme is verified through simulation experiments.Compared with linear controller,sliding mode controller and Backstepping BLF controller,the designed controller has active defense capability.It can tolerate longer attacks in the presence of Do S attacks and weaken the impact of attacks on system performance.