"TU Ilmenau presents the technology of tomorrow at HANNOVER MESSE." As slogans go, this one definitely sounds more like the proud pragmatist presenting hard-won achievements than the persuasive PR people peddling the purported eighth wonder of the world. What’s more, by zeroing in on autonomous driving, the scientists at TU Ilmenau have also picked out a topic that will actually affect everyone in some way, and is proving somewhat controversial. If you're the kind of person who can't help wondering "what if..?", you'd be well advised to head along to the pavilion. "What if a self-driving car hits an elderly pedestrian because it mistakes an advertising hoarding that is blown into the road for a child and serves to miss it?" you ask? That's exactly the kind of fear that the team from TU Ilmenau wants to allay. The model system they are presenting at HANNOVER MESSE 2018 clearly sets out how their concept of autonomous driving can rule out dramatic chains of events precisely like this.

The team from the "Simulation and Optimal Processes Group" led by Professor Pu Li are using their exhibit to show how high-tech model cars can reliably master even tricky maneuvers like maintaining a set distance from other vehicles or immediately detecting obstacles. What's more, these impressive technical achievements are based on self-learning algorithms and don’t require any human intervention during driving. The team's approach to tackling specific tasks is based on a strategy that is calculated online, making it safer and less prone to faults than conventional processes. To achieve their aims, the researchers developed an optimization procedure that takes equal account of environmental conditions, vehicle dynamics and the preferences of the people traveling in the vehicle. The concept can be easily transferred to full-scale vehicles and should therefore be of great interest to the automobile industry. In fact, just last year, the development won the team from TU Ilmenau the Audi Autonomous Driving Cup, an international competition aimed at delivering fully automated driving functions and the associated software architectures.