Effect of Pavement Markings on Machine Vision Used in ADAS Functions

Advanced Driver Assistance Systems (ADAS) like Lane Departure Warning (LDW) and Lane Centering Assistance (LCA) have experienced low customer acceptance and market penetration despite the technology being in active development for several years now. This trend is attributed to the inability of many of the perception systems to consistently detect lane markings and localize the vehicle on roads with poor lane markings, changing weather conditions, and occlusions. Currently, no standards or benchmarks are available to evaluate the quality of either the lane markings or the perception algorithms. This work seeks to establish a reference framework that can be used by transportation agencies to evaluate the effect of pavement markings on ADAS functions. Previous works have looked at these problems using on-road pavement markings. However, environmental factors like sun glare, shadows, and road illumination, affect Lane Detection (LD) performance with changing roads and driving directions. It is necessary to eliminate external environmental factors that may affect the study to selectively evaluate the effect of pavement marking material characteristics on LD performance. An extensive video dataset was developed by driving on a closed-course road with customized pavement markings (having different material properties) under controlled conditions. The data were collected at different times of day and weather conditions and evaluated on state-of-the-art lane detection algorithms. An Analysis of Variances (ANOVA) statistical model was employed to analyze the effect of pavement marking material characteristics and evaluation conditions on LD performance. A systems approach was presented by correlating the algorithm performance data to the environmental factors, lane marking types, color, material, and the retroreflectivity of pavement markings. Key recommendations are listed to allow transportation agencies to understand the benefits and impacts of modifying pavement marking practices.