On the Way to Level 5

Modern driver assist systems, like autonomous cruise control and parking assistance, can intervene in steering, acceleration and deceleration even today. Some test vehicles are already in command of the next stage: highly automated driving. But it will still take a number of years before we will see full automation and cars without a steering wheel.

The way to autonomous driving will pass through several levels. At the lowest level (“Level 0”), drivers are not assisted at all – they must steer, brake and accelerate themselves. Things are made a little easier for them at Level 1 with assist systems, like adaptive cruise control (ACC), which maintains a constant distance from the vehicle ahead without the driver having to use the brake and accelerator pedal all the time. Level 2 provides even more convenience: here, parking assist systems actively intervene in the steering. This is state-of-the-art in today’s production vehicles.

The next level (Level 3) is highly automated driving: Here, the driver no longer needs to keep permanent attention on the vehicle. It can perform many actions itself, such as change lane. In the meantime, the human being can be getting on with other things, but must be able to take over control again within 10 to a maximum of 30 seconds according to current international discussion.

Various test vehicles show what these systems will be able to do one day in the future. For example, IAV has been traveling the approximately 70 kilometer journey from Chemnitz to Dresden Airport on a highly autonomous basis in a converted VW Golf for a number of years now. There is hardly any call for the driver any more – he or she only has to take action at a number of traffic lights because the system cannot reliably match up the traffic light signals to the lanes. To do this, the vehicle would need to get further information via a carto-infrastructure network.

As soon as vehicles have reached Level 4 – fully automated driving – drivers will hardly need to worry about the traffic situation at all. The vehicle will then control almost all functions and maneuvers independently and the early warning time for human action could be about a minute. IAV’s test vehicle provides this level of automation too – at the moment, though, only on a test track.

The maximum level of automation is reached with autonomous driving, this being “Level 5”. These vehicles instantly give away their autonomy: they have neither a steering wheel nor pedals – no human intervention whatsoever is envisaged any more. These vehicles are no longer any optimizations of conventional models but developed from scratch for autonomous driving.

But there are still a number of hurdles to overcome on the path to the driverless vehicle. Although the Vienna Convention on Road Traffic has already been amended so as not to stand in the way of increasing automation any more, it is now a matter of answering other questions: what will the procedure look like for approving such vehicles for volume production? And is there any broad acceptance for them in society?

The technology still also needs to be advanced. To replace the driver, the surroundings sensor system must provide a 360° image of the traffic situation. For this, a redundant sensor concept is necessary which, for example, combines the data from a video camera and a LIDAR sensor. But it will also be important for the vehicle to be connected with its environment, such as traffic lights. Therefore as a member of the Car2Car consortium, IAV is involved in the development of standards for Car2X communication.

The map material will be the key to success of autonomous driving. On the base of this material the algorithms compute appropriate driving strategies. Google has a head start, based on its own map service, which is now to be countered by the alliance of German manufacturers Audi, BMW and Daimler through the acquisition of the “Here” map service with the associated access this will provide to high-quality data. However, further efforts are still necessary to improve the data. At the moment there are still irregularities in the content of information between adjacent map sections which have to be rectified since they can lead to erratic behavior on the part of the vehicle and to undefined driving dynamic.

Basically, there are hardly any technical differences on the path to autonomous driving between passenger cars and trucks, and level classification is identical, too. However, the business cases differ. In the case of platooning, for example, trucks drive close up to each other to save fuel. For this, the driver at the head of the platoon would have to be compensated because he is the only one not to benefit from the slipstream the others are driving in.