基于传感器融合的户外道路下无人驾驶卡车姿态估计方法研究

Research on Attitude Estimation Method of Autonomous Truck Under Outdoor Road Based on Sensor Fusion

论文针对户外道路下的无人驾驶卡车,提出了一种基于微机电系统惯性测量单元和激光雷达融合的姿态估计方法。论文的主要内容包括三个方面:1)基于扩展卡尔曼滤波器设计无加速度数学模型,克服惯性测量单元的纵向和横向测量漂移。2)从激光雷达一维点云分布建立多线扫描几何模型,分析路面障碍物(包括凸起障碍物和凹陷障碍物)的几何特征得到道路特征。3)结合激光雷达提供的道路特征,利用所提的无加速度模型,移除因户外不平整道路导致惯性测量单元出现测量滑移的现象。实验结果表明该方法能够有效减少惯性测量单元的因外部加速度导致的测量漂移现象,精确地提供姿态估计。

This paper focuses on multi-axis autonomous truck,proposes a vehicle attitude estimation method under outdoor road based on sensor fusion of MEMS(Micro-electromechanical Systems)IMU(Inertial Measurement Unit)and LiDAR.The main contents of this paper are threefold.Acceleration-free kinematic model is designed by means of Extend Kalman Filter to overcome MEMS IMU's measurement drift along longitudinal and lateral directions.Multi-line scan geometrical model is built from 1D point cloud distribution,geometrical feature of negative(pits)and positive(speed bumps)obstacle of road surface are analyzed to supply distance and size information of obstacles.Combined with the obstacle information provided by the LiDAR,the proposed accelera⁃tion-free model is used to remove the inaccurate measurement of MEMS IMU caused by the outdoor uneven road.The experimental results show that the proposed method can effectively reduce the measurement drift of the IMU and accurately provide attitude esti⁃mation.