In real-life road tests, the actions of other road users cannot be clearly defined. Cameras mounted on the vehicle roof are therefore used to monitor the actual traffic situation.

Yet, as Schittenhelm knows, it isn’t possible to scientifically record every factor precisely. “Attentive drivers are also subconsciously aware of a host of incidental factors. These include things like how the other driver might react. Is he or she phoning, for example, and thus possibly distracted? Unfortunately, even the best sensors in the world can’t register this sixth sense.” For this reason, mini-cameras in the driving simulator continuously record images of the test driver’s legs, feet, posture, and face. Likewise, a special gaze-tracking system not only documents the driver’s head and eye movements but also analyzes them in real time. Then it’s time for the various teams to evaluate the results and decide which technical systems might be of help to the driver in such situations.

Promising ideas for intersection assist systems are first replicated in the simulator and tested further in order to establish their effectiveness. “When programming this kind of virtual system, we can define how and when it recognizes the danger, whether it uses a shrill noise or a purple flashing light to warn the driver, and what kind of interventional action it will take should this prove necessary,” explains Schittenhelm. “What we specifically don’t want to do is take away the driver’s own power of decision,” he emphasizes. “Our aim is to ensure that drivers are warned of potential danger as early as possible in order to give them sufficient time to react and, if necessary, help them take evasive action.”

Gaining time

Some problems, however, seem intractable at first. Take the issue of time, for example. Just how quickly can a driver-assist system respond to an impending crash? Often there are only one or two seconds between recognizing the danger and the actual collision. “In a really critical situation, there’s very little time left for the driver to react between the warning signal and the impact. A fraction of a second can make all the difference,” Schittenhelm explains. “It’s therefore vital that sensors respond as quickly as possible.”

Naturally, this raises new questions. Can a sensitive, ultra-responsive electronic system reliably distinguish genuinely hazardous situations from harmless ones? Frank is familiar with this quandary: “Too many false alarms end up achieving the exact opposite: the driver either doesn’t take the system seriously or switches it off altogether.”

Reducing accident risk

However, as tests with the driving simulator demonstrate, research teams at Daimler are coming to grips with this problem as well. When one third of test drivers were given no electronic assist system whatsoever, one third a brake assist, and one third a brake assist plus an intersection assist, it was clearly shown that such technology substantially reduces the risk of an accident.

Meanwhile, these new warning systems are now undergoing tests in the real world. At the same time, the teams involved in these areas of research have become part of the Accident Data Analysis Project Center. “We put a batch of test drivers in cars, get them to take specific routes through traffic with the help of a specially prepared navigation and information system, and then observe their reactions on the basis of the same parameters as in the simulator,” Frank explains. “Our ultimate aim is to develop an effective intersection assist system for everyday use,” Schittenhelm adds. “It should be able to reliably identify hazardous situations, provide sufficient warning, and if necessary inter-vene and prevent an accident.”