Joining traffic made easy

New technologies can support the driver by taking over parts of the driving. However, at the moment these systems are not well accepted. One reason for this is that many systems do not take into account the respective traffic situation. Psychologists and engineers from the Virtual Institute "Human Automation" want to change this. Travelling along the right-hand lane of the motorway. Automatic Cruise Control (ACC) maintains a desired speed and keeps a safe distance to the car ahead. In many situations this is a comfortable option for the driver. But as the traffic becomes heavier, and cars keep cutting in, the system brakes, annoying the driver. "It's at moments like these that drivers switch off the essentially useful aid, because it does not respond to the traffic situation and reduce the distance," says Prof. Mark Vollrath, explaining his criticism. This is why Vollrath and colleagues from the DLR Institute of Transportation Systems in Braunschweig are researching more intelligent, adaptive systems. Psychologists and engineers from the DLR joined forces with the Institute of Automotive Engineering at the RWTH Aachen University and the Interdisciplinary Centre for Traffic Sciences Würzburg to establish the Virtual Institute on "Human Automation in Traffic". Initially, says coordinator Mark Vollrath, they focused solely on just one traffic situation: entering the motorway and filtering into the oncoming traffic. It's not as complex as city traffic, but much more difficult than situations that available systems already manage. First of all, traffic psychologists have to find out when drivers become stressed and respond incorrectly. On the one hand, they analyse the driver and driving behaviour of test subjects in specific situations in the driving simulator. And, on the other, they analyse behaviour in real-life traffic with the research vehicle ViewCar. The first results confirm that the scientists are right to develop situation adaptable systems. Because the mistakes that drivers make depend on the level of traffic.

At low traffic levels, drivers become careless. This is why a system must give only minimum support and warnings in this situation so that they do not become even more careless. At medium traffic intensities, drivers make the most mistakes. They have to keep their distance to the vehicle ahead, while, at the same time, keeping the traffic next to them in sight and mind, recognising the proper gap to merge, and then braking or accelerating as necessary. In this situation, assistance systems which take over parts of the driving may be helpful. At very high traffic levels, cars drive so slowly that enough gaps appear. In qualitative terms, the results are no surprise. "However, we can now also quantify at what objectively-measurable traffic level the system has to react," explains Mark Vollrath. Hence, he and his colleagues are able to design a system that initially adapts to the various traffic situations. In the next step, they aim to enable the system to even predict the driver's behaviour on the basis of gaze direction in order to prevent unnecessary actions and warnings occurring.

The engineers involved in the collaborative project are developing a method to automatically measure the three different traffic levels. It uses laser and radar scanners attached to the research vehicle ViewCar. At the same time, psychologists are working with the driving simulator to study which assistance systems are feasible and best support the driver. No decision has been reached yet as whether a filtering-in assistance system should, in certain situations, automatically brake or accelerate, or whether the system should only show the driver the proper gap to take, or whether it should, plainly and simply, warn the driver if the situation becomes precarious. The filtering-in systems developed at the DLR are to be installed in a car that will be able and allowed to travel in real traffic in three years' time. But Mark Vollrath is already thinking one step ahead. He believes it is possible for cars of the future to drive autonomously - including joining traffic automatically and speed regulation. "But until that happens, plenty still remains to be done with the assistance systems that usefully support drivers in the various situations."