实验室条件下道路交通信号场景用例采集可行性研究

Feasibility Study on Acquiring Scene Cases of Road Traffic Signals in the Condition of Laboratory

道路交通信号的准确识别关乎自动驾驶汽车的安全运行。开放或封闭道路条件下道路交通信号场景用例的参数不可控,实验室条件下可以灵活控制其各项参数,但限于实验室场地大小限制等因素,无法拍摄获得远距离的交通信号。本文通过理论分析和实地验证焦距、视场大小及拍摄距离的数学关系,提出了一种在实验室场地受限的条件下采用不同焦距镜头模拟不同距离下拍摄采集道路交通信号场景用例的方法,同时在实验室条件下改变道路交通信号参数并模拟不同距离进行场景用例采集。该方法能够满足实验室内灵活调节交通信号参数、快速获取大量场景用例的要求,丰富自动驾驶汽车对道路交通信号识别测试的内容。

Road traffic signals have a great influence on the safe operation of autonomous vehicles. The parameters of road traffic signals are uncontrollable on open or blocked roads, while parameters can be adjusted flexibly in the laboratory. However, shooting and acquiring the distant road traffic signals is limited by the size of laboratory. This paper proposes an approach to use camera lens with different focal lengths to simulate the acquisition of road traffic signals at different distances. Meanwhile, scene cases are acquired by adjusting parameters of road traffic signals and simulating different distances in the laboratory. The proposed approach can meet the requirements of adjusting parameters of road traffic signals flexibly and acquiring numerous scene cases quickly. Also, it can enrich the content of recognition and measurement of road traffic signals which aims at autonomous vehicles.