Autonomous Driving Oriented Model Predictive Control With Guaranteed Vehicle Steering Stability

Vehicle intelligence is one of the key issues and core technologies in the current and future development of automobiles.As the level of vehicle intelligence increases,the level of autonomous driving gradually transitions from no automation to fully autonomous driving.At the stages of human-machine cooperative control and fully autonomous driving,different steering stability issues arise due to human and road factors under regular and extreme operating conditions.Therefore,researching steering stability control methods for multiple operating conditions is of significant importance for ensuring the safety of autonomous driving vehicles.As the level of autonomous driving increases,the impact of human-machine interaction and the safety requirements under multiple operating conditions pose new challenges to steering stability control methods.These challenges can be summarized as follows:(1)Under regular operating conditions,different trust levels of human drivers in human-machine cooperative control may lead to conflicts and affect steering stability.Thus,addressing conflicts and avoiding adverse effects on steering stability under regular operating conditions is a problem to be solved.(2)Under extreme operating conditions,rapid steering maneuvers at high speeds may cause tire slip,and human drivers may have difficulty making correct decisions.It is indeed a key issue to address how to control the vehicle’s unstable state and ensure steering stability,especially in extreme operating conditions where it may exacerbate vehicle sideslips or even lead to rollovers.Therefore,controlling the vehicle’s instability and ensuring steering stability under extreme operating conditions are urgent issues.(3)In multiple operating conditions consist of both regular and extreme operating situations,designing control methods to ensure steering stability and driving safety is a crucial problem in achieving fully autonomous driving.This thesis focuses on addressing these issues.Firstly,the human model and the vehicle model for autonomous driving are established and regular operating condition human-related factors and extreme operating condition vehicle-roadrelated factors that influence steering stability are analyzed.This forms the foundation for designing steering stability control methods for autonomous driving under multiple operating conditions.A 3-degree-of-freedom nonlinear model and a simplified 2-degree-of-freedom linear model for the vehicle system and calculate the equilibrium points of the vehicle system.The open-loop stability using phase-plane and Lyapunov methods is analyzed,highlighting extreme operating condition vehicle-road-related factors that affect steering stability.The driving data is collected from human drivers and survey questionnaires through field experiments.Based on the driving data,a driver model is establised that simulates steering control behavior of human drivers with different skill levels.Using survey questionnaires and statistical methods,the relationship is analyzed between driver trust in autonomous driving systems and differences in driving skills,identifying regular operating condition human-related factors that affect steering stability.Secondly,to address the steering stability control problem affected by human-related factors under regular operating conditions,a model predictive control(MPC)approach is proposed based on the stability region for human-machine cooperative steering control.A trajectory tracking control method is designed based on linear MPC that considers human-machine cooperative weighting,incorporating stability region constraints to ensure steering stability.Using the analysis results of the human-machine trust relationship under regular operating conditions,a fuzzy logic-based human-machine cooperative weighting is designed.The effectiveness of the proposed methods is validated in reducing the physical burden on drivers,mitigating humanmachine conflicts,ensuring steering stability and driving safety through vehicle dynamics simulation software.Next,to address the steering stability control problem affected by vehicle-road-related factors under extreme operating conditions,a MPC approach is proposed based on unstable equilibrium points for autonomous steering under extreme operating conditions.A quasi-infinite horizon nonlinear model predictive control(NMPC)to drive the vehicle from an unstable state to the neighborhood of an unstable equilibrium point.Terminal element constraints are designed to guarantee asymptotic stability of the system and use the Koopman operator theory-based MPC fast-solving method to handle nonlinear dynamics-induced computational burden efficiently.The non-convex optimization problem is converted into a quadratic programming problem with terminal constraints,achieving an average solution time less than the sampling time.Comparative simulation experiments validate the effectiveness of the proposed methods in controlling vehicle instability under extreme operating conditions.Finally,to ensure steering stability and driving safety in multiple operating conditions of autonomous driving,a NMPC approach is propose for path following considering multiple equilibrium points.The path following problem involves multiple equilibrium points,which is transformed into a regulation problem based on a single equilibrium point.A quasi-infinite horizon NMPC method is designed,which is proved to be feasible and convergent of path following with parameterized paths.Comparative simulation experiments verify the effectiveness of the proposed methods in ensuring steering stability and driving safety under multiple operating conditions.In summary,this thesis researches MPC methods for steering stability in autonomous driving under multiple conditions,providing viable solutions to address steering stability issues at different stages and under multiple operating conditions in autonomous driving.