FOSS@Porsche
Research @ Porsche Digital Campus
18th December 2023 · 3min reading time
The Porsche Digital Campus is an innovative initiative with the goal of bringing together research and real-world application in the fields of AI, machine learning, and data science.
Tell us a little more about yourself, your role at Porsche Digital and what topics drive you.
We are the research team and part of the Porsche Digital Campus: An
international group of software technology experts with a clear focus on the
application of artificial intelligence as well quantum computing.
Porsche Digital has published several research projects in the field of artificial intelligence on GitHub. How would you explain these projects to a five-year-old? What problem do these projects solve for Porsche?
For the first project, “Using deep learning to construct stochastic local
search SAT solvers with performance bounds”, imagine you have a tricky
puzzle. Computers have a special friend, like a magical helper, that helps
them solve these puzzles faster and better! We built a magical helper that
is really smart and can learn the best way to solve more difficult puzzles
quickly. So, with this helper, computers can play and finish more tough
puzzles in less time! It’s like having a super-smart friend to help you with
your games!
For the second project, “Distribution and volume based scoring for Isolation
Forest’, imagine you have a super-smart robot that can find which toys are
different from the others. We made this robot even smarter! The team taught it a
new way to look at all the toys and find the different ones more accurately. We
tested the new smart robot with lots of toys and found out it can spot the
different ones even better than before! So, it’s like having an upgraded robot
friend who’s really good at finding unique toys!
For the third project, “FARSEC: A Reproducible Framework for Automatic Real-Time
Vehicle Speed Estimation Using Traffic Cameras”, imagine you have a smart camera
that can watch cars and see how fast they are going. This is important to help
cars move smoothly on the road and to keep everyone safe. Some smart cameras are
good but can only work well with certain videos. In this project, we developed a
new software algorithm for smart cameras that can work well with many different
videos and can understand how fast cars are going even if the camera is moving.
This new camera is easy to use and can be made even smarter in the future!
By publishing AI research projects as open-source projects, you’ve made a big step ahead in our FOSS movement. What was your initial motivation to take this path? Can you tell us why you’ve chosen to release your project on GitHub?
Today, providing the code and data necessary to reproduce the results
presented in a research paper is a necessary requirement for scientific
credibility. So being able to make our code available alongside our
publications is absolute key to establishing that our work is as credible as
that of academic institutions. Porsche’s official GitHub organization
is the perfect place to do so.
What did you learn along the way? What were the benefits of this strategy for you, your team and Porsche Digital?
We were very keen to see the results of the research. Knowing that we would
make the code public was exciting and the team paid even more attention to
writing clean code, keeping the size of the repo manageable, and having
sensible code structures. This, again, improved the quality of the code and
hence the progress of the whole project.
With a glimpse into the future: What are your further plans regarding FOSS Contributions? Who would you like to see next joining our FOSS Contributor community?
Publishing on GitHub will become the new default way of making the code for
studies available alongside academic publications. So, keep your eyes
peeled!
