Abstract

Classification of vehicles into distinct groups is critical for a number of applications including freight and commodity flow modeling, pavement management and design, tolling, air quality monitoring, and intelligent transportation systems. The main goal of this paper is to demonstrate how data from Light Detection and Ranging (LiDAR) sensors could be leveraged to distinguish between specific types of truck trailers beyond what is generally possible by the traditional vehicle classification sensors (e.g., piezoelectric sensors and inductive loop detectors). A multi-array LiDAR sensor enables the construction of 3D-profiles of vehicles since it measures the distance to the object reflecting its emitted light. This paper shows how point-cloud data from a 16-beam LiDAR sensor are processed to extract useful information and features to classify semi-trailer trucks hauling ten different types of trailers: a reefer and non-reefer dry van, 20 ft and 40 ft intermodal containers, 40 ft reefer intermodal container, platforms, tanks, car transporter, open-top van/dump and aggregated other types (i.e., livestock, logging). K-Nearest Neighbors (KNN), Multilayer Perceptron (MLP), Adaptive Boosting (AdaBoost.M2), and Support Vector Machines (SVM) supervised machine learning algorithms are implemented on the field data collected on a freeway segment that includes over seven-thousand trucks. The results show that with the SVM model, different trailer body types can be distinguished with a very high level of accuracy ranging from 85% to 98%.

摘要

将车辆分成不同的组对于许多应用至关重要，包括货运和商品流建模、路面管理和设计、收费、空气质量监测和智能交通系统。本文的主要目标是展示如何利用来自光探测和测距 (LiDAR) 传感器的数据来区分特定类型的卡车拖车，这超出了传统车辆分类传感器（例如压电传感器和电感回路探测器）。多阵列 LiDAR 传感器能够构建车辆的 3D 轮廓，因为它可以测量到反射其发射光的物体的距离。本文展示了如何处理来自 16 光束 LiDAR 传感器的点云数据以提取有用的信息和特征，以对牵引十种不同类型拖车的半挂车进行分类：冷藏和非冷藏干货车、20 英尺和 40 英尺多式联运集装箱、40 英尺冷藏多式联运集装箱、平台、坦克、汽车运输车、敞篷货车/自卸货车和其他类型的聚合（即牲畜、伐木）。K-最近邻 (KNN)、多层感知器 (MLP)、自适应提升 (AdaBoost.M2) 和支持向量机 (SVM) 监督机器学习算法在高速公路路段收集的现场数据上实施，其中包括超过七千卡车。结果表明，使用 SVM 模型，可以以 85% 到 98% 的非常高的准确度区分不同的拖车车身类型。