Using mathematics and statistics to make self-driving cars more reliable

Lukas Sommeregger’s research focuses on developing methods to determine the lifespan of computer chips. The doctoral student, who also works at Infineon, recently received the Infineon Innovation Award.

Today, more than 1,000 semiconductor chips are incorporated in a car. Self-driving cars, expected to dominate our roads in the future, will need an even greater number of these electronic components. If the vehicles are deployed more intensely than previously, even around the clock, it is likely that the chips will be depleted more quickly. Consequently, a semiconductor manufacturer like Infineon, which develops and produces chips for vehicles of this kind, is faced with the following crucial question: How can we make sure that the chips will outlast the lifetime of the car?

Lukas Sommeregger is an applied mathematician and statistician and is employed as an industrial PhD at Infineon in Villach. As part of his doctoral thesis, he is developing models that can be used to calculate how many of the chips are likely to reach their performance limits over time. But what data serves as the basis for this? Lukas Sommeregger explains: “You can’t just drive cars around on a trial basis for twenty years and then check how many chips develop defects over time. Instead, we use the so-called lifetime stress test: This means creating conditions that are more extreme than those found under normal circumstances, and the cars switch back and forth between them. This stress test can involve switching between very hot and very cold conditions, or operating in very high humidity. This yields data based on the use of around 70 chips during the stress test.” However, since millions of these components will be on the road in practice, we also need statistics in order to be able to draw meaningful conclusions from the data. Lukas Sommeregger uses the data obtained to create models that can be used to calculate the lifespan of chips. The ultimate goal is to enable self-driving cars to attend to their own repairs proactively at some point in the future: “Ideally, the vehicle would sense that at a given time it would be advisable to drive itself to the workshop and get a chip replaced.”

“Being able to solve problems that exist in real life” brings Lukas Sommeregger a lot of joy. The PhD position at Infineon allows him to conduct applied research while remaining “fully hands-on” with the subject matter. He hopes to remain in this field. Lukas Sommeregger first completed the Bachelor’s degree programme in Technical Mathematics and then followed this with the Master’s degree programme in Mathematics, both at the University of Klagenfurt. He went straight on to pursue a doctorate, under the supervision of Jürgen Pilz (Department of Statistics). When asked whether mathematics has always been easy for him, we learn: “Yes, but I must add that it also depends a lot on how the subject is taught in school. If you receive a strong foundation, you can successfully make the transition to university mathematics. At university, more creativity comes into play. You start with a problem and there are many ways to solve it.” Lukas Sommeregger was recently awarded the Austrian Infineon Innovation Award for the “Best PhD Topic” in recognition of his work, which is already well advanced.