Abstract

Research reveals that visual information forms the major portion of the received data for driving. At night -owing to the, sometimes scarcity, sometime inhomogeneity of light- the human physiology and psychology experiences a dramatic alteration. It is found that although the likelihood of accident occurrence is higher during the day due to heavier traffic, the most fatal accidents still occur during night time. How can road safety be improved in limited lighting conditions using DMD-based high resolution headlamps? DMD-based pixel light systems, utilizing HID and LED light sources, are able to address hundreds of thousands of pixels individually. Using camera information, this capability allows 'glare-free' light distributions that perfectly adapt to the needs of all road users. What really enables these systems to stand out however, is their on-road image projection capability. This projection functionality may be used in co-operation with other driver assistance systems as an assist feature for the projection of navigation data, warning signs, car status information etc. Since contrast sensitivity constitutes a decisive measure of the human visual function, here is then a core question: what distributions of luminance in the projection space produce highly visible on-road image projections? This work seeks to address this question. Responses on sets of differently illuminated projections are collected from a group of participants and later interpreted using statistical data obtained using a luminance camera. Some aspects regarding the correlation between contrast ratio, symbol form and attention capture are also discussed.

Abstract

The introduction of light emitting diodes (LED) in automotive exterior lighting systems provides opportunities to develop viable alternatives to conventional communication and sensing technologies. Most of the advanced driver-assist and autonomous vehicle technologies are based on Radio Detection and Ranging (RADAR) or Light Detection and Ranging (LiDAR) systems that use radio frequency or laser signals, respectively. While reliable and real-time information on vehicle speeds is critical for traffic operations management and autonomous vehicles safety, RADAR or LiDAR systems have some deficiencies especially in curved road scenarios where the incidence angle is rapidly varying. In this paper, we propose a novel speed estimation system so-called the Visible Light Detection and Ranging (ViLDAR) that builds upon sensing visible light variation of the vehicle's headlamp. We determine the accuracy of the proposed speed estimator in straight and curved road scenarios. We further present how the algorithm design parameters and the channel noise level affect the speed estimation accuracy. For wide incidence angles, the simulation results show that the ViLDAR outperforms RADAR/LiDAR systems in both straight and curved road scenarios. Publisher's Version