Interview: A (big data) hackathon for fresh ideas
For the first time, Porsche organized a public hackathon on the topic of big data and artificial intelligence. Thirty participants from universities, from Porsche and from partner companies were selected from over 90 applicants. These included statisticians from the Ludwigs-Maximilians-Universität in Munich and young machine learning researchers from KIT, as well as international Porsche employees from Prague and colleagues from MHP management and IT consultants.
In this interview, the two initiators Bettina Frana (Chassis Innovation & Strategy, Porsche AG) and Tobias Grosse-Puppendahl (Innovation Management Electrical/Electronic Engineering, Porsche AG) look back on the event’s successful format.
Bettina, Tobias — how did you come up with the idea of organizing a hackathon?
In the last few years, hackathons have proven to be an excellent medium for finding solutions to interesting company challenges in a focused and creative way. At Porsche we have come across many of these topics; for example, determining the coefficient of friction between the road and the tires or identifying traffic situations for validating autonomic driving situations. To get some fresh ideas for this challenge, we organized the hackathon.
Your focus is on big data. From your point of view, what does this term mean?
Primarily, big-data just means large amounts of data. We are increasingly seeing these amounts of data in vehicles; on current platforms several gigabytes of data are processed every second, for example. For our hackathon we used a somewhat smaller dataset that is based upon CAN recordings from 25 vehicles. Over a period of six months, we recorded several terabytes of data that we then used for the hackathon.
Why did you focus upon big data and artificial intelligence?
In our day to day work with our colleagues it has become evident that the topics of big data and the application of artificial intelligence methods represent enormous potential for Porsche. By analyzing these volumes of data, errors can quickly be found and even predicted. What’s more, sensors can be simulated to open up whole new application scenarios. Last but not least, safe autonomous driving can only be achieved with precise methods for validating the driving functions, which operate on large datasets.
What tasks do the participants have to tackle as part of a Porsche hackathon?
For our big data hackathon, we prepared two challenges in the Porsche Digital Lab: Our colleagues working on the chassis are developing a method of determining the coefficient of friction (grip) between the road and the tires — however, sensors are very expensive and impractical, so they are pursuing other parameters for deriving the coefficient of friction, such as driving behavior and weather. As such, they are developing a kind of virtual sensor. As we have already recorded a “ground truth” — values from a reference sensor — using all of the vehicles in our dataset, we can simply compare these with the digital predictions of the coefficient of friction. On the other side of things, our colleagues in electronics are currently dealing with the validation of assistance functions, especially for autonomous driving. They are developing a shadow mode system, which drives virtually in the background but does not intervene in the driving process. The potential actions of the shadow mode system can then be compared to the actual actions of the driver. In this way we can judge whether a new function works safely or if errors arise in certain circumstances. This dataset can be used to identify appropriate driving situations, known as “corner cases”.
What do you expect from the participants?
We hope that the hackathon participants will be able to provide valuable food for thought for our challenges; ideally, we want to learn from them and use them for our developments at Porsche. It is particularly important for us that we also maintain good contacts with the external participants, so that we can identify talent and inspire them to apply for a position at Porsche, for example. We also wish to strengthen our cooperation with our partner companies and within Porsche, so that colleagues can form a better network.
Are you planning to organize more events like this?
We will primarily be treating this first Porsche Big Data Hackathon as a pilot project from which we can learn about holding a hackathon in an automotive environment. The positive feedback and keen interest we received have given us great momentum, so we can imagine organizing events like this regularly.
An interview with the two hackathon-initiators Bettina Frana (Chassis Innovation & Strategy, Porsche AG) and Tobias Grosse-Puppendahl (Innovation Management Electrical/Electronic Engineering, Porsche AG).
More about inspiring women and men and many other technologies can be found on our Porsche Medium Blog or Twitter and Instagram.
