In recent years,automated driving technology has achieved very rapid development.It is currently in a transitional period from L2 assisted driving to L3 or above automated driving.The primary purpose of developing autonomous driving is to reduce traffic accidents caused by driver errors.Therefore,safety is the core attribute of automated driving.The realization of automated driving is inseparable from the support of complex software and hardware.How to avoid harm to drivers and other traffic participants due to the failure of software and hardware(that is,to ensure functional safety)is a key issue that needs to be solved urgently.ISO 26262 is an international standard for functional safety in the automotive field.This standard is based on V model and provides design guidance for the development of automotive products in the concept phase,system phase,and software and hardware phases.In June 2020,the United Nations Economic Commission for Europe(UNECE)promulgated the first approval regulation for L3 Automated Lane Keeping System(ALKS),which set requirements on the function specification,operational domain and safety design of the ALKS system.L3 is the first step from assisted driving to automated driving.However,there are few studies on the functional safety of automatic driving system at L3 level or above.Besides,lateral control is the last guarantee of the automated driving system;its importance is self-evident.Unfortunately,most of the current lateral control research focuses on the realization of ontology function,and seldom involves the safety design after the failure of ontology function.In response to above problems,based on ISO 26262 standard and UNECE regulation,this thesis will conduct the following research on the lateral control of the ALKS system:(1)Concept phase analysis and development of ALKS system.Firstly,the ALKS functional description and operating design conditions(including the three aspects of the human-vehicle-environment: the external environment,the state of the vehicle itself,and the state of the drivers and passengers)will be defined;and then,the initial architecture of the ALKS system will be proposed as the basis for subsequent safety analysis.Based on the functional description of the ALKS system,Hazard and Operability Analysis(HAZOP)is used to identify malfunctions,and then the vehicle operation scenarios are considered to form vehicle-level hazardous events,and the Severiy(S),Exposure(E)and Controllability(C)are dcomprehensively utilized to evaluate Automotive Safety Integrity Level(ASIL)of the above hazardous events and finally the corresponding Safety Goals are obtained.(2)Decomposition of safety requirements based on fault tree.Firstly,the violation of the Safety Goals(Hazardous Events)related to lateral control are assigned as the top event of Fault Tree Analysis(FTA),and the initial architecture elements of the ALKS system are used to decompose the top event and allocated by functional safety requirements.Secondly,the functional safety requirements and the initial system architecture are combined to form the functional safety architecture(FSA)of the ALKS system,which includes the physical view of the system using the control structure diagram of the Systems Theoretic Process Analysis(STPA)and the system behavior view expressed by the state machine diagram.Finally,based on the logical "Input-Processing-Output"(IPO)of functional safety architecture,functional safety requirements are decomposed into software safety requirements(SSR)for lateral control modules,which serves as the basis for the following lateral control safety design and verification.(3)Safety design and verification of ALKS lateral control function.Based on the generated software safety requirements for lateral control modules,with the goal of achieving "FailOperational" for L3 ALKS,the functional safety design of lateral control is carried out.Firstly,based on the software safety requirements and system functional safety architecture,the software architecture of the lateral control module is further refined as the basis for safety design.Secondly,the two sets of control algorithms for the realization of heterogeneous redundancy(Model predictive control: with the goal of accuracy optimization;Pure tracking control: with the goal of safety optimization)are further refined.Thirdly,the specific lateral control functional safety logic that uses above two sets of algorithms to achieve heterogeneous redundancy and safety monitoring is explained.Finally,Simulink/Carsim co-simulation is used to implement above safety design,and then,by injecting typical lateral control faults(unintended steering and lost steering),above safety design is verified.The verification results show that the safety design meets the corresponding software safety requirements and the realization of safety goals are guaranteed. |