基于激光反射强度特征的智能汽车路面估计方法

Road Estimation Method for Intelligent Vehicles Based on Laser Intensity Distribution Feature Analysis

摘要:

为了提升智能汽车行驶安全性，适应高速、大侧偏等极限工况应用场景，轮胎-路面附着系数峰值估计技术越来越受到汽车主动安全控制领域研究的关注。提出了一种基于激光雷达的路面估计方法：基于极大似然估计方法求解了结构化道路常见类型路面的激光雷达反射强度分布模型参数，并依此建立典型路面数据库；利用Kullback-Leibler散度表征反射强度分布相似度，结合所建立的路面数据库辨识路面类型，然后映射出对应的峰值附着系数估计值。试验结果表明，提出的轮胎-路面附着系数峰值估计方法的准确率达到90%以上，能够灵敏地检测出路面突变现象，且对白天和夜晚不同的光照条件具有鲁棒性。

Abstract:

In order to improve the driving safety and adaptability of intelligent vehicles to application scenarios under extreme conditions such as high speed and large lateral slipping state， the technology of tire-road peak adhesion coefficient （TRPAC） estimation （hereinafter referred to as road estimation） has aroused more and more attention in the field of active safety control. A road estimation method based on lidar is proposed. Based on theory of maximum likelihood， laser intensity distribution model parameters of several typical kinds of structured road surfaces are solved， based on which， the database of typical road surfaces is established. By using Kullback-Leibler Divergence （KLD） to represent the similarity of intensity distributions， road surfaces can be classified according to the established database， and the estimated value of TRPAC can then be mapped. The experiment results show that the proposed method can estimate TRPAC with an accuracy of more than 90%， that the sudden change of road condition can be sensitively detected， and that the proposed method has robustness to different illumination conditions in the day and at night.