Despite great technical progress in vehicle safety, according to the WHO approximately 1.2 million fatalities occur on the world’s roads every year. Thus great efforts are undertaken to reduce the number of road fatalities and serious accidents, or at least to mitigate their impact on road users. The introduction of environment perception based Advanced Driver Assistance Systems (ADAS) in road vehicles is expected to improve traffic safety significantly. In today’s vehicles, the prevalent ADAS focus primarily on the longitudinal driving direction, e.g. Autonomous Emergency Braking (AEB) systems and Adaptive Cruise Control. Whilst the functional range of these systems continues to expand, there exists a very large portion of critical vehicle crashes which are not addressed, mainly caused by vehicles leaving the roadway laterally.
Today, Electronic Stability Control (ESC) is one of the only established Active Safety Systems covering emergency situations in the lateral direction, and always dealing within the physical limits of vehicle dynamics. Despite the very high effectiveness of ESC systems, there remain many cases in which it is not possible to prevent unintended lateral roadway departures, especially in cases of driver drowsiness or inattention, e.g. on American highways or European rural roads. Preventing roadway departure crashes, which cannot be covered by today’s lateral guidance/lane keeping ADAS, is motivation behind developing a system for road departure protection. Road Departure Protection systems expand today’s lateral ADAS by a ctive road keeping in emergency situations before reaching the physical limits of vehicle dynamics. Based on environment perception means like road edge detection or road course preview, the system should actively intervene when unintentionally leaving the roadway. By automated vehicle control, the system keeps the vehicle on the roadway, thereby protecting against roll-over accidents or collisions with roadside obstacles or oncoming traffic.
This paper presents the roadway departure problem we face today via accident data and different use cases and gives insight into the state of the art Active Safety functions and research activities. The functionalities to address the selected use cases will be described, including functional architectures, a road edge detection algorithm, approaches to preview the course of the road, sensor fusion concepts, a function cascade, and activation strategies. First test data will illustrate the function and working area of a Road Departure Protection system.
A reliable and real-time capable perception algorithm will be demonstrated. For this algorithm, different image processing techniques are applied to mono camera images to estimate the parameters of a geometric model of the road edge. It works without any supplementary knowledge about the road infrastructure.
Two system architectures will be presented, which differ in the used surrounding sensors, actuators, functional capabilities, and system cost. One variant uses a radar sensor, stereo camera and an Electric Power Steerin g (EPS) system, whereas another solution uses a mono camera and an ESC system. The vehicle steering capabilities and limitations of the ESC based steering are discussed in the paper. Finally, an outlook to future work and possible extensions will be given.