Modern headlamps adapt in real time to changing driving situations. The magic name behind this dynamic lighting is “sensor fusion”.
Driving in the dark requires increased levels of concentration. Drivers may be dazzled either by other vehicles or may themselves dazzle other drivers. And visibility is impaired as the light beam strays off the road when the vehicle goes around corners or over crests. Modern headlamps with adaptive lighting provide the solution. They adjust the beam of light in a fraction of a second to suit the situation.
To do this, the headlamps need to know the precise position of the vehicle and its surroundings. And this knowledge comes from sensors that use cameras, radar and invisible infrared light to detect the environment. Also coming into the mix is vehicle data from the gas pedal, steering wheel and shock absorbers as well as road geometry data from the navigation system. “The fusion of all of this data is the only way to create a detailed model of the vehicle’s environment and the vehicle’s motion,” says Dr. Maximilian Austerer, Head of Technical Business Development at the OSRAM Continental joint venture. “This information is sent to the high-resolution headlamp systems, which then change the distribution of light to suit the specific situation.” For example, they can mask out the beam of light in the area of an approaching or preceding vehicle in order not to dazzle the driver.
Relatively complex
But fusing this data is not as easy as it sounds. The process behind it is quite complex. The sensor signals are sent to a data fusion block. This analyzes the relevant data and uses the results to create a highly precise environmental model containing the position of the vehicle and the relative positions of other objects. The next step makes use of algorithms to generate a prediction of the environment model. “That’s not an easy task, because with two vehicles each moving at 60 mph, their relative position changes by more than 160 feet in just one second,” adds Austerer. That's why accurate predictions are so important. This information is then transmitted to the scene creation module which generates the high-resolution light form for the projection system.
Glare-free beam over the bumps
The particular challenge here is to take into account latency times, in other words the delay between a signal being detected and the system responding to that signal. “We need to bring together the information from many different sensors. But we have the additional problem that the sensors deliver the data at different times,” says Austerer. “We have to use sophisticated algorithms to compensate for the different latency times.”
This complex process works even when driving over a bumpy road. “Let's say we're driving over a crest in the road. Normally, the light/dark boundary created by the headlamps will move up and down. Thanks to sensor information from a gyro or from the compression and extension of the suspension springs, our algorithms can make predictions about changes in the road in a split second and compensate for the movement of the light beam.”
Lightening the load on the driver
To make the most reliable predictions, OSRAM Continental also uses learning systems. If several drivers manually switch to low beam at a specific point on the road – a construction site or the entrance to a town, for example – this information is stored in the cloud and conveyed to the on-board systems of other vehicles via the connectivity module. When those vehicles reach that section of the road, the light assist system sends an automatic low-beam command to the headlamps. “This is all about detecting standardized driving situations and initiating intelligent responses to relieve the pressure on drivers and improve road safety. The information collected by the vehicle is evaluated and networked to improve the knowledge base for all vehicles,” explains Austerer. Ultimately, sensor fusion provides a new user experience: Stabilized beams, highly dynamic dark zones and glare-free high beam make driving at night a relaxing experience.