The project AVRM (autonomous vehicles and road markings) aimed to examine how vehicles’ advanced driver assistance systems (ADAS) are constructed, how they function and how they detect road markings on the Nordic road network. Focus was on the systems lane departure warning (LDW) and lane keeping assist (LKA).
Both the literature study and the interview study concluded that if the human eye can detect the road marking, then the road marking is machine-readable. However, only a few studies had been conducted in wet conditions relating machine-readability to road marking functionality.The pilot study aimed to test equipment and to find a method to connect machine-readability data with contrast ratio under various weather and light conditions, and to reveal possible problems before conduction of a main study. Results from both the literature study and the pilot study pointed out that wear and lack of road markings were the parameters related to road markings per se that contributed to poor machine-readability.
The analysis showed that in daylight, there was no strong relationship between machine-readability and conventional road marking performance parameters. In addition, machine-readability was higher on multilane roads (99%) compared to on two-lane roads (93%), which may be explained for example by fewer curves on larger roads. Although data showed that machine-readability of broken lines was somewhat worse than that of solid lines of line width 0.1 m, this could be an effect of factors related to the (minor) roads where broken lines with 0.1 m width are commonly used.
In sum, there are many factors unrelated to road markings that influence machine-readability. There are no clear relationships between machine-readability and conventional performance parameters. It should also be kept in mind that since retroreflectivity is a parameter measuring the performance in night-time, it could not be expected to affect daylight readability. As long as the road markings are visible for the human eye, they can be expected to be machine-readable as well.