By Michael Hamm
Burghard Böttcher, ADAC
Report of ADAC Symposium Glare in Road Traffic
On 25 and 26 March 2025, the ADAC organized a symposium on glare in road traffic to discuss the factors that cause glare in road traffic. Several contributors from motorist groups, ophthalmologists, and researchers presented their findings.
Together with motorist clubs from Austria, Switzerland, Belgium, Luxemburg, and Great Britain, ADAC conducted a survey of motorists at the end of 2023. In Germany, 1,089 driving licence holders were surveyed. The questions were more subjective without situation-based context, as shown in the table below: “Do you feel dazzled by the lights of other vehicles?” The answers showed that glare in road traffic affects a large proportion of road users. Among other questions, around three-quarters of all respondents were in favour of adopting legislation so that there is less glare.
As was shown in the ADAC survey, 80 per cent of the respondents claim they are sometimes, regularly, or almost always glared.
ADAC concludes from the survey answers that maximum values may be necessary for illuminance in the main light cone as well as maximum luminance values. Adaptive lighting systems would contribute to less glare in road traffic. Effective headlight cleaning systems should be provided for LED headlights regardless of their light output.
Dr. Michael Hamm, TU Darmstadt
Glare Contributors – A Metastudy on Scientific Research about Different Aspects of Glare
This was an overview about various research on glare of studies that were only published as PhD thesis, or specific research reports. 2004 the Darmstadt Laboratory of Lighting Technology published a report showing that with higher surrounding luminance, e.g. provided by brighter headlamp illumination, visual performance triples. A second investigation showed that visual performance while driving is decreasing. Compared to standstill, while driving at 60 km/h, visual acuity decreased by 21 per cent. So, results found under laboratory conditions might be not fully correct compared to reality.
Another PhD Study revealed that a similar online internet questionnaire gave already 2014 the results that 79.6% of the surveyed persons said they were “sometimes, often or always” glared. Glare thus is no new effect of the latest headlamp development. About 3 to 5 per cent of all times the topology from the street is not flat, so glare can occur as shown in the example below.
Christian Hinterwälder, Audi
Headlamp Misalignment: Cause and Effects on glare.
The presentation gave an overview about contributors on glare. Contributors include aiming accuracy, stick-slip effect, tire pressure variation, temperature load, fuel tank fill level, and many more. A study was shown that a small deviation of aim, of about 0.5% (plus or minus) did not deliver more glare, because the cutoff is still below oncoming eyes, but resulted in less visible range.
Aiming accuracy seems to be a very delicate topic. Human eyes have about a factor of 3 to 4 higher variation (i.e. tolerance) than digital aiming devices which use gradient evaluation of a video picture. Especially the human eye combined with a Fresnel-based aiming device shows the biggest deviations.
Ernst-Olaf Rosenhahn, Marelli Germany
Headlamp Misalignment: Cause and Effects on glare.
The presented statistics about headlamp glare taken under real traffic conditions for about 4,500 opposing situations during nighttime driving show a percentage of 9.8% with a glare effect with oncoming driver’s eyes below the cutoff line inside the active light distribution. The reasons include road topography, asymmetric part of the low beam above the horizon, misaim, car inclination by load, and often a combination of these reasons. These glare events can be distinguished in two glare levels called “moderate glare” and “strong glare”. Out of the 9.8 % of all glare events the strong glare events happen for 1.1 % of all oncoming situations.
The majority is in the category of moderate glare. Anyhow, the focus of glare reduction should focus on the strong glare events, which are about 1% of all opposing situations. An estimate was given using road topography. Driving a crest or a bridge plus a curve, creates a glare pulse in a distance between the cars from about 120 m down to 60 m, in addition to the basic glare level. It is caused by a vertical curve and has a maximum glare illuminance of about 5 lx at 72 m distance, and an exposure of 3,4 lx·s, which is a level clearly above the limit and this creates strong glare for oncoming drivers.
Anil Erkan, Audi
Glare Causes in Nighttime Traffic – How can we minimize the risk of glaring headlights to increase traffic safety?
The presentation identified glare causes that occur during the product life cycle of a vehicle. These causes should then be assigned to the different phases in the product life cycle (production, operation) and evaluated according to their relevance. Basic contributors to glare are headlight adjustment, headlamp cleaning, levelling (automatic dynamic) and adaptive functions. Focusing on ADB adaptive lighting functions, a study was presented. The results of the studies show that ADB systems achieve detection ranges comparable to high beams (120 to 130m), while they lie within the range of low beams for both disability glare and discomfort glare (de Boer 7.1 to 7.6).
The realization of an adaptive driving beam is currently only possible in high beam mode, as there is no legal basis in low beam mode. But the implementation of an adaptive low beam could further reduce the glare risk in addition to ADB, through fully adaptive adjustment of the low beam distribution. Some examples were given: adaptation of the cutoff line for oncoming cars, adaptation to uphill/downhill situations, dynamic adaptation of light intensities especially in foreground. There were many good arguments for adaptive low beam.
Korbinian Kunst, TU Darmstadt
Headlamp Performance Ratings: A Comparative Analysis of HSPR and VLPS
During the previous ISALs the headlamp evaluation system HSPR has been presented. The China New Car Assessment Program (C-NCAP) follows a similar approach through its Vehicle Lighting Performance Score (VLPS), applying structured evaluation criteria to assess headlight effectiveness in real-world driving conditions while integrating simulation-based testing methods. Both HSPR and C-NCAP VLPS evaluate headlight performance based on key parameters, including straight and curved lane illumination, pedestrian visibility, intersection lighting, and glare calculation
But there are clear differences between the systems. Some have minor impact (e.g. street width, changed measuring zones A, B), flux for low beam area, live measurements on reaction time / response, bonus points for automatic levelling, bonus points for correct aiming.
VLPS limits the maximum score: low beam and high beam systems are capped at 10 points, and ADB systems are also capped at 10 points. This approach means that a very good low/high beam system without ADB can achieve the same maximum 10 points as an advanced ADB system. In HSPR, by comparison, a vehicle equipped with an ADB system would naturally score higher due to better illumination and dynamic adaptation, encouraging lighting innovation. The capping of the scoring points is also demonstrated by calculating the score at Point A for the high beam assessment. Even if a system offers a greater viewing distance for the driver, its performance is no longer rewarded beyond a certain threshold. In the future, ADB systems will play an increasing role not only in rural areas but also in urban and semi-urban environments. New scoring systems and real-world use cases must be developed to properly assess and incentivize the next generation of intelligent, adaptive lighting functions.
Elisabeth Kemmler, TU Darmstadt
Discomfort Glare from LED Lighting: Impact of the Size of the Glare Light Source and the Background Luminance
A laboratory study with a LED glare light source was carried out. The used luminaires produce luminances up to approximately 17,000 cd/m2 but could be dimmed to thirteen different luminance levels. The size of the illuminated area was regulated by using three aperture panels with different aperture sizes. A screen behind the glare light source was homogeneously illuminated to generate three different background luminances in the mesopic area during the experiment: 0.1 cd/m2, 0.35 cd/m2 and 1.0 cd/m2.
The study consisted of five parts: a formal part to clarify organizational matters, an adaptation phase, a training round, the main experiment and a closing feedback round. The glare stimuli were presented in randomized order. The duration of each stimulus was 3 sec. There was no time limit for the evaluation, but the subjects were advised to submit the glare rating as quickly as possible.
The results show that there is a logarithmic relationship between the luminance of the glare light source and the experienced magnitude of discomfort glare. Regarding the influence of the size of the glare light source on the glare perception, it was found that there was not a single case, where the average de Boer rating of a light condition was lower for the smaller source.
It can be assumed that disability glare has an influence on the discomfort glare perception. At higher background luminances, smaller contrast differences must be processed by the visual system. In other words, the optical nerves are less irritated, which leads to more visual comfort. The conclusion from this experiment indicates there is a logarithmic relationship between the luminance of the glare light source and the experienced magnitude of discomfort glare, but all depending on the ambient luminance.
Sinan Yargeldi, Mercedes-Benz
Optimization of ADB systems for perceptibility when using high resolution light source modules
The presentation focused on non-lighting effects while optimizing a ADB system. focusing on how system perceptibility impact visibility, safety and detection accuracy. All calculation in ADB systems can be separated into three steps: information generation, pixel pattern generation, and image generation and projection. Step 1 is mostly done by the ADAS/object detection, outside of the headlamp. The presentation focused on the technical processing of information and process through the ADB system. The system design influences the system latency, the internal bit depth in the computational pipeline, the light module refresh rate and the light module resolution.
The angular speed of an incoming car is proportional to the curve radius and the vehicle distance. This means that bigger safety margins must be added for higher system latencies. Thus, the usable effective resolution is severely decreased in fringe situations, even for delay times as low as 50ms (leading to a shift of 1.35° at 20m vehicle distance). That means the deglaring area must be bigger as it looks in a standstill photo.
The findings indicate that higher bit depth, resolution and frame rates can enhance safety as well as the detection accuracy and response time for both human drivers and assistance systems. Additionally, the system latency is shown to be a huge bottleneck for the performance of high-resolution ADB systems, needing to be countered by a fitting architecture and software design.
Additionally, it is shown that handling more than 30 deglaring areas is not necessary for most driving scenarios. Therefore, we propose the usage of convex quadrilaterals as deglaring areas, offering the best compromise between computational complexity and illumination. Depending on the available power, the number can be increased starting with a minimum of 10 areas.
Ignacio Cadenas, Renault
PWM Frequencies – Feasibility to avoid their negative effects
A literature study was the base for the presentation. Negative effects for human vision (flicker, stroboscopic effect and phantom array) should be avoided. A long year practice of OEM was a minimum standard that guarantees safety by removing the principal human visual effects, such as consistent flickering.
Unfortunately, cameras are impacted by PWM. This means that any PWM frequency used in exterior lights is at risk of being seen/detected by any kind of camera, if that frequency is not multiple of the video fps applied.
EE components were presented that are able to dramatically increase the PWM frequency like Buck or Boost Converters. The markets seem not to be ready for dramatic changes.
Only OEMs have the necessary levers in each one of the affected aspects, and mainly OEMs would benefit from the outcome. The presentation therefore mentioned the need to drive this discussion across all industry, which would represent a clear compromise with the underlaying safety aspect of the topic.
Hyeran Kang, Yeungnam University
Effect of Rear Lamp Shape and PWM Frequency on the Visibility of the Phantom Array Effect
This presentation investigated how the shape of rear lamps (horizontal vs. vertical configurations) affects the visibility of the phantom array effect. Using two production car lamps—a horizontally elongated lamp and a vertically elongated lamp some experiments were done. The threshold frequencies at which the phantom array effect became perceptible were investigated. Subjective ratings were collected. The results suggest that rear lamp geometry can affect the perceptibility of the phantom array effect.
Rear lamp geometry interacts with the dynamics of eye movements during driving. Since saccadic eye movements are predominantly horizontal, a vertically narrow light source is more likely to produce abrupt temporal sampling, leading to enhanced temporal aliasing. This may intensify the appearance of visual artefacts, such as the phantom array effect. In contrast, a horizontally elongated lamp may allow for greater spatial-temporal overlap during eye movements, reducing the visibility of the effect.
