面向稀疏路网车辆行驶安全的瞬态纵向最优化识别区域模型研究

Study onvehicle safety in sparse road network based on instant longitudinal optimized detection model

针对现有机器视觉测距技术测量距离不足和测量精度较低不能运用稀疏路网工况的问题,提出一种基于瞬态纵向最优化识别区域的模型.首先,依据麋鹿实验分析了两种常见发散态纵向识别区域模型,并建立了瞬态纵向最优化识别理论模型;其次,基于Carsim进行了复杂路况制动距离多样化分析,并由修正安全制动距离计算出最优化识别区域模型对应的稳态成像焦距值.实验证明,在30~100m测量范围内的平均相对误差低于3.5%,绝对误差小于2m.

In order to solve the problem that traditional distance detection method based on computer vision,which couldn＇t be applied in sparse road network condition since its low accuracy.A novel transient distance detection model was proposed in this paper.Firstly,two kinds of common emanative distance detection models were analyzed based on Elk test and a theoretical distance detection model was proposed,then multi complex road and vehicle conditions were analyzed based on Carsim simulation,finally stable focal length was obtained which corresponds to the ideal distance detection mode.The experimental results shown that the average absolute and relative detection error were less than 2 meters and 3.5% in the area from 30 meter to 100 meter.