In times of increasingly more powerful driver assistant systems, automatic driving appears on the horizon of many automotive R&D managers. At the Innovation Summit in Munich, BMW explained the pros and cons of automatic driving. The discussion highlights the complicated splits between the feasible and the desirable.
The perspective of automatic driving is real, explained BMW R&D manager Peter Waldmann at the innovation Summit, a high-level discussion forum in Munich. According to the BMW expert, automatic driving has a high potential to improve the safety situation on highways and streets. "Will cars drive autonomously in 20 years?" Waldmann asked. The answer was a clear "I don't know" – not technology alone needs to be considered.
Waldmann sketched the vision of an "electronic co-pilot" which takes control of the vehicle during annoying traffic situations such as congestions or stop-and-go traffic. In critical situations, the electronic co-pilot warns the driver and provides instructions to defuse the situation; in quickly developing dangerous situation it intervenes directly and thus helps to avoid serious accidents. "It is important that the driver under all circumstances can override the system", Waldmann said.
Most components of the technology required for automatic driving are available. Such a system embraces three functional components: Vehicle control, position determination and environment detection. The first group, vehicle control systems, already features a high degree of maturity. BMW demoed a trial in which a serial production car drove lap on the Nürburgring racing track at high speed – with a driver who never touched the steering wheel. "What makes this racing track difficult for automatic driving is its country road character", Waldmann explained; tight curves and a narrow road complicate algorithms and set high demands to the speed of the control loops.
In order to implement such a system, sensor data fusion is a precondition, Waldmann explained. In the case at hand, the system had to process GPS data, camera-based lane recognition data as well as vehicle-internal data – plus the data of a highly precise digital map of the area. The Nürburgring trial has been conducted in October 2010.
Another application for automatic driving is what BMW calls an "emergency stop assistant system". In the case a driver encounters sudden health problems such as a heart attack – not exactly an exotic case in times of an ageing population – the emergency stop assistant steps in, stabilizes the vehicle within its lane and then drives autonomously to the hard shoulder – of course considering the traffic situation. Finally, it makes an emergency call, providing position and other relevant data to the emergency services.
In terms of technology, the emergency stop assistant is a rather complex data processing device: Since it must avoid any collision with other vehicles, it needs to be equipped with rather sophisticated environment detection. BMW has implemented such an assistant system that combines the input signals of a laser scanner as well as video cameras, radar and ultrasonic sensors. Again, a case of sensor fusion.
Before the industry integrates such assistant systems into their set of wheels, a number of issues needs to be cleared, Waldmann said. For instance, a reliable environment recognition along with equally reliable situation interpretation algorithms must be available. At the same time, a comprehensive digital map with very high accuracy is required.
And, perhaps the most critical issue: The legal situation for automatic driving need to be set. This includes a number of non-engineering topics, such as liability in the case of a malfunction.
Waldmann made clear that at least the technology issues are solvable. However, beyond technology and legal situation there might be other obstacles for the deployment of automatic driving systems: Acceptance. "Where is the fun of driving?" commented a listener.
This article originally appeared on EE Times Europe.