面向智能汽车测试的弱势群体服饰色彩研究

智能汽车测试是其技术开发与应用中必不可少的环节,封闭场景下测试目标物准确反映真实道路环境下交通对象特性是保障测评结果可信的关键,而道路弱势群体服饰色彩是相应测试目标设计的关键参数,也是智能车测评相关标准中要求的一个主要指标。为此,通过对中国某省份2018～2020年间重大交通安全事故案例的分析和筛查,得出178例弱势道路使用者群体伤亡人员样本,首先提取样本服饰颜色,然后选取适当的色彩空间,将色彩数据从RGB(Red-Green-Blue)空间转换至LUV(Lightness-Chroma)空间。以转换结果作为聚类参数,采用K-means聚类算法,获取受害者样本基于季节、出行方式等不同因素下的服饰代表颜色。区别现阶段欧洲标准中目标物黑色上衣/蓝色长裤的搭配组合,黑色上衣/黑色长裤作用于符合中国国情的自动驾驶场景中测试目标物的服饰颜色更具代表性。鉴于中国新车评价规程(China-New Car Assessment Programme, C-NCAP)选取行人目标物与自行车骑行者目标物,将目标物服饰改为黑色上衣/黑色长裤组合,以测试目标物与测试车辆位置分别构建相对横向及纵向运动的多个场景,...

Clothing Color of Vulnerable Groups for Intelligent Vehicle Testing

The testing of intelligent vehicles is paramount to their technological development and application. Whether the test surrogates can accurately reflect the characteristics of the traffic objects in the actual road environment of in-field testing is key to ensuring the credibility of the evaluation results. Meanwhile, the clothing color of vulnerable groups on the road is a key parameter for designing the vulnerable group surrogates as well as a main indicator in the relevant standards for intelligent vehicle evaluation. By analyzing major traffic accident in one representative province in China from 2018 to 2020, a sample of 178 victim cases is obtained. Firstly, the clothing color of the samples is extracted. Subsequently, the appropriate color space is converted from RGB (Red-Green-Blue) space to LUV (Brightness, Chroma) space. Using the conversion result as the clustering parameter, the K-means clustering algorithm is applied to obtain the representative clothing color based on different factors such as age, season, and travel mode. Different from the clothing color combination of a black tops/blue pants of the surrogates in the current European standard, a black tops and black pants combination is more representative of the scenarios in China. To conform to China-New Car Assessment Programme (C-NCAP) regulations, multiple Near and Far scenarios with black tops and black pants of pedestrian surrogate and bicyclist surrogate are constructed respectively. The collision points between the testing vehicle and surrogate at 25%, 50% and 75% of the bumper of the testing vehicle in the corresponding scenarios are analyzed to evaluate the response ability of the intelligent vehicle equipped with an automatic emergency braking system. The results show that in all scenarios, the testing vehicle can successfully identify the target and brake actively. These tests verified the feasibility and effectiveness of the black tops/black pants combination under the current testing standards. The results provide sufficient data support for intelligent vehicle testing, improve relevant standards and regulations in the transportation industry, and promote the development of intelligent vehicle testing technology.