Browse Publications Technical Papers 2018-01-1079

Study on Test Scenarios of Environment Perception System under Rear-End Collision Risk 2018-01-1079

The foundation of both advanced driving assistance system(ADAS) and automated driving (AD) is an accurate environment perception system(EPS). However, evaluation and test method of EPS are seldom studied. In this paper, naturalistic driving environment was studied and test scenarios for EPS under rear-end collision risk were proposed accordingly. To describe driving environment, a new concept named environment perception element(EPE) was first proposed in this paper, which refers to all the objects that the EPS must perceive during driving. Typical environment perception elements include weather and light conditions, road features, road markings, traffic signs, traffic lights, other vehicles, pedal cyclists and pedestrians and others. Driving behaviors collected in Shanghai, China were classified and rear-end collision risk scenarios were obtained and described using EPEs. Probability distribution of EPEs was therefore obtained. Afterwards, the correlation between EPEs and risk level of scenarios (evaluated by maximum longitudinal deceleration) were revealed by means of Fisher’s Exact Test. Based on these two characteristics of driving environment, typical test scenarios for EPS were established with the help of cluster analysis, and the test scenarios were simulated in Prescan. These test scenarios were generally consistent with the probability distribution of EPEs, making the test results reliable. Results from this paper fill the gap between the high demand of dynamic and representative EPS test scenarios and the existing static picture database used in development and validation of EPS and are of great significance to the development of ADAS and AD in China.


Subscribers can view annotate, and download all of SAE's content. Learn More »


Members save up to 16% off list price.
Login to see discount.
Special Offer: Download multiple Technical Papers each year? TechSelect is a cost-effective subscription option to select and download 12-100 full-text Technical Papers per year. Find more information here.