汽车巡航系统前方关键目标识别研究

Research on Key Target Distinguishing Method of Automobile Cruise System

针对巡航系统在弯道行驶路段常出现目标车混乱或丢失的现象,应用扩展卡尔曼滤波理论进行巡航系统前方弯道关键目标识别研究。建立纵向、侧向与横摆的非线性三自由度车辆模型,设计了横摆角速度的扩展卡尔曼滤波器,实现了道路曲率的在线实时估计。建立了弯道关键目标识别模型并给定了关键目标判定依据。采用模糊控制理论,设计了多目标车换道模糊逻辑控制器,预判危险换道车辆,确定了直道行驶的关键目标车。利用MATLAB/Simulink搭建立了三自由度车辆模型的扩展卡尔曼滤波器仿真模块,采用双移线输入实现了车辆横摆角速度与质心侧偏角估计的验证。在Carsim中构建车辆模型、传感器模型及道路模型等仿真环境,实现了与Simulink控制模块的联合仿真,验证了弯道多目标行驶及直道行驶工况下关键目标的准确识别,结果表明:采用扩展卡尔曼滤波理论的弯道目标识别方法能够准确判定弯道关键目标,有效避免系统在弯道路段因目标混乱或丢失而造成的追尾事故。基于模糊控制理论的关键目标模糊判别方法可预判直道行驶危险换道车辆并准确识别关键目标。

Adopting the extend Kalman filter theory, a key target distinguishing method is researched in order to avoid the goal confusion or missing of automobile cruise control system （ACC） on the curve road. An extend Kalman filter on yaw rate and a key target distinguishing model are established to estimate the road curvature in real time and decide the critical target based on a nonlinear three-DOF vehicle model including the longitudinal, lateral and yaw prejudge direction the dange Adopting the fuzzy control theory, a multi-object lane-changing controller is presented to r lane-changing objects and distinguish out the key target on the straight road. A simulation model is elaborated with MATLAB/Simulink to validate the estimation of yaw rate and side slip angle under high- speed double lane input conditions. Simulation experiments are given for multiple targets in the curve road by adopting Carsim and MATLAB/Simulink co-simulation under building vehicle, sensor and road models in the Carsim conditions. The simulation results show that the key target distinguishing methods can identify accurately the key target, and avoid effectively rear-end collisions owing to the goal confusion or missing on the curve road, and prejudge the danger lane-changing objects on the straight road.