news

       Automation of vehicles is often cited as one of the most disruptive technologies, both economically and environmentally. But for this to actually happen, the surrounding infrastructure must adapt to support the efficiency and safety of driverless cars. Connected infrastructure is critical to the design and certification process, as well as the test environments that need to be deployed. In this context, the connection between infrastructure and vehicles is also critical to ensuring operating conditions during a transition period that could last for decades.
       In addition to the automotive and C-ITS aspects of traffic management for autonomous vehicles, another important issue that must not be overlooked is the roads themselves and their readiness for driver assistance technologies. Harald Mosböck, Vice President Road Marking Systems Europe and APMEA at Swarco and Vice President of the European Road Federation (ERF), is well aware of the role of road markings in both human and machine vision. When asked whether our roads are ready for driverless cars, he replies: “Not yet.”
       Currently, lane markings are read by two types of sensors: cameras and lidars. Even the most modern vehicles equipped with high-definition maps need such sensors to record lane markings and road signs and to determine whether all the data is truly comparable and consistent with each other. The fact is that there is still no uniformity in the markings and road signs on the roads. The differences are too great. “Improving the uniformity is a prerequisite for increasing the reliability of the readings of the different types of sensors,” explains Mossböck.
       At the international level, the improvement and standardization of road markings is mainly carried out by standards organizations in Japan, the United States and Europe. The latest edition of the North American Manual of Uniform Traffic Control Devices (MUTCD) requires that the main road markings be 150 mm (6 inches) wide and more uniform, which is a major drawback in the implementation of CAV.
       Most recently, a major standards body proposed to the Federal Highway Administration that future highway networks not only have standardized lane widths, but also require dashed lines at exit ramps and uniformly indicate the width of gaps for broken lines. Work is currently underway to improve retroreflectivity, a complex topic. Another challenge is how to improve lane markings in road construction zones so that they can be reliably and consistently read by autonomous or assisted driving machines.
       Referring to leading European actions to improve road markings, including those initiated by industry associations such as the European Road Federation, Harald Mosböck alluded to the European Commission’s requirement in the Third Road Vehicles Package that “Member States must ensure that road markings and road signs are properly designed and maintained so that they can be easily and reliably recognised by both human drivers and vehicles equipped with driver assistance systems or higher levels of automation.”
       In 2019, the European Council decided to revise the Road Safety Management Directive and tasked the European Commission with setting up an expert group to examine the development of road markings and road signs that would make them easier to recognise by vehicles and, of course, by people. The expert group began its work, but the entire process was delayed due to the epidemic. The results were initially expected to be announced in mid-2021, but will now only be published in spring 2022. “Overall, we can say that communication between car manufacturers, industry associations and Brussels on the quality requirements for road markings has improved significantly,” comments Mosbock.
       The proposal put forward by the European Road Research Foundation (ERF) is very similar to the trend in the United States. The formula was developed with the human eye in mind and is intended to improve road safety. Interestingly, numerous studies have shown that the factors that contribute to safe and comfortable driving for the human eye are also applicable to machine vision. The “150 x 150 formula” assumes a line width of 150 mm and a retroreflectivity (RL) > 150 mcd/m²/lx in dry conditions (RL > 35 mcd/m²/lx in wet conditions).
       To get more information on how well cameras and lidars read lane markings, Swarco, together with its Austrian partner ZKW, a leader in automotive lighting technology, conducted tests in the world’s largest climate tunnel in Vienna.
       We installed eight different road marking systems, from standard to high-performance, read by three different cameras and eight different lidar sensors. We simulated 24 different weather conditions, including perfect day and night, wind, storm, rain and fog.
       The test results also confirm a rule of thumb from Mobileye, the world leader in automotive camera technology: Information that a camera can read, the human eye can read. Information that the human eye can read may not be readable by a camera.
       “On the one hand, cameras are strong in their ability to discern contrast. On the other hand, cameras are not good at reading glare from oncoming vehicles. As for lidar sensors, some highly reflective road markings certainly improve lidar readability and detectability,” says Mossbock. “In addition, lidar can very accurately read orange road markings, even with special near-infrared pigments. This is critical given the ongoing debate in the US about whether lane markings in future road construction zones should be orange. It turns out that lidar is not affected by glare. So when cameras and lidar are used together in a car, the result is excellent readability and reliability for driver assistance.”