This paper proposes an improved autonomous emergency braking (AEB) algorithm intended for intelligent vehicle. Featuring a combination with the estimation of road adhesion coefficient, the proposed approach takes into account the performance of electronic hydraulic brake. In order for the accurate yet fast estimate of road ahead adhesion coefficient, the expectation maximization framework is applied depending on the reflectivity of ground extracted by multiple beams lidar in four major steps, which are the rough extraction of ground points based on 3σ criterion, the accurate extraction of ground points through principal component analysis (PCA), the main distribution characteristics of ground as extracted using the expectation maximum method (EM) and the estimation of road adhesion coefficient via joint probability. In order to describe the performance of EHB, the response characteristics, as well as the forward and adverse models of both braking pressure and acceleration are obtained. Then, with two typical roads including single homogeneous road and fragment pavement, the safe distance of improved AEB is modeled. To validate the algorithm developed in this paper, various tests have been conducted. According to the test results, the reflectivity of laser point cloud is effective in estimating the road adhesion coefficient. Moreover, considering the performance of EHB system, the improved AEB algorithm is deemed more consistent with the practicalities.

本文提出了一种针对智能车辆的改进的自主紧急制动（AEB）算法。该方法结合道路附着系数的估计，考虑了电子液压制动器的性能。为了准确而快速地估计前方道路附着系数，根据多波束激光雷达提取的地面反射率应用期望最大化框架，主要分为四个步骤，即基于 3 σ的地面点粗提取。标准，通过主成分分析（PCA）准确提取地面点，使用期望最大值法（EM）提取的地面主要分布特征和通过联合概率估计道路附着系数。为了描述EHB的性能，获得了制动压力和加速度的响应特性以及正向和反向模型。然后，以单一均质路面和碎石路面两种典型道路为基础，对改进型AEB的安全距离进行建模。为了验证本文开发的算法，进行了各种测试。根据测试结果，激光点​​云的反射率可以有效地估计道路附着系数。此外，考虑到 EHB 系统的性能，