Cooperative Adaptive Cruise Control Of Connected And Autonomous Vehicle With Uncertain Preceptions

Cooperative Adaptive Cruise Control(CACC)can use advanced on-board sensors,controllers,actuators and communication equipment to coordinate and efficiently coordinate the longitudinal behavior of each vehicle in the platoon comprehensively.It has great potential to improve the maximum capacity of road traffic,prevent the spread of traffic shocks,and improve the safety and fuel economy of road traffic systems.It is an important scenario and direction for the practical application of intelligent networked vehicle technology.This technology is rooted in advanced perception and communication technology,and has certain reliance on the perception and real-time communication capabilities of platoon.The existing CACC algorithm is mainly based on the assumption of deterministic perception,that is,the sensor can provide accurate perception data.However,sensor errors have existed all the time and are difficult to eliminate.The difference between the above assumptions and the actual situation can easily lead to degradation and failure of CACC performance,and lead to some problems such as reduce road traffic capacity,or improve the risk of traffic accident and fuel efficiency.In response to the above research gap,this theis based on stochastic optimal control proposes a CACC method(Cooperative Adaptive Cruise Control with Uncertain Preceptions,CACC-UP)of two-preceding following information flow topology oriented to the uncertain perception and control efficiency.According to dynamic programming,a real-time solution for the optimal control of the leader of platoon under variable speed is proposed.Combined with the Separation Theorem and the Lyapunov stability analysis,the stability of the controller is verified from a theoretical perspective.The works of this thesis are as follows:(1)Based on the four-element framwork,this thesis uses second-order node dynamics,two predecessor-following information flow topology and constant headway strategy to construct a CACC-UP node model.Based on the comprehensive analysis of the various uncertainties that may appear in the state and measurement,the normal disturbance is used to describe the uncertainty of the state and the measurement.In terms of measurement uncertainty,this paper uses the optimal weight method to initially fuse the perception data from multiple sensors,thereby initially reducing the perception uncertainty.(2)Considering multiple factors such as car following safety,ride comfort,etc.,this thesis designs a CACC-UP controller based on stochastic optimal control.The control problem is described as a linear quadratic Gaussian problem that minimizes the error of the distance and the relative speed of the two vehicles,and the smallest change in acceleration.According to the Separation Theorem,this thesis decouples into the solution of the deterministic system and the real-time estimation of the Kalman filter.Aiming at the problem that the open-loop input is not considered in the existing deterministic system solution method,this thesis uses the dynamic programming to construct the gain of the controller quantity and the iterative form of the gain based on the optimality principle and the stochastic Bellman equation.The step-adaptive Kalman filter estimates the optimal state estimator at each moment,thereby establishing a realtime solution method for the optimal control of the leading vehicle under the condition of shifting.(3)This thesis analyzes the closed-loop stability of the designed CACC-UP controller.Also based on the Separation Theorem,the stability of the controller is decoupled into the stability of quadratic Gaussian problem and the Kalman filter.The Lyapunov direct discriminant method is used to prove the closed-loop stability of the quadratic Gaussian problem.The stability of the Kalman filter is confirmed by the proof of the uniform controllability and uniform observability of the system.The above analysis process not only proves the stability of the system,but also provides a certain basis for the selection of controller parameters.(4)In order to verify the performance of the CACC-UP controller,different scenarios such as initial disturbance convergence,leading vehicle speed change,natural driving pilot vehicle and emergency braking safety test were designed.MATLAB was used to simulate and verify the performance of the controller under different noises and sensitivity analysis.The simulation results show that the CACC-UP controller can still ensure the stability of the platoon and traffic flow,and it has a restraining effect on different disturbances in the presence of large uncertainties in the existing sensing and control equipment.