RESEARCHERS USE SUPERCOMPUTERS TO TRAIN SMARTHINKING CARS
Are self-driving cars cool or just plain scary? We have all heard the hype around AI reality hitting the roads, but what are the boffins doing to make sure they’re ready for our dynamic and ever-changing driving environments? A team over at the University of Wuppertal are using some serious supercomputing power to get the AI behind our future cars revved up for success!
Dr. Matthias Rottmann and his crew are teaching machines to figure out those unpredictable and unfamiliar road scenarios human drivers navigate with ease. To prepare our future self-driving vehicles for the mysterious and the unknown, the team uses the JUWELS Booster supercomputer at the Jülich Supercomputing Centre. With its high-performance computing (HPC) resources, this ultimate driving teacher is giving AI-governed systems the upgrade they need for making the right calls.
The secret sauce behind this high-tech teaching approach? Active learning. This life-changing machine learning method helps train applications over many iterations, getting the program to flag what it’s unsure of and requesting feedback to guide its learning. It’s like the smarter way to teach, as opposed to making the AI eat up thousands of labelled images like it’s binging on TV shows.
So, is AI really ready to jump into driving us around without all the nitty-gritty training? Rottmann and his go-getter group are making leaps and bounds on not just working with these bad boys, but getting us ready for more AI conquering endeavors. This work is about seeing how we can use the training tools and approaches for other applications besides autonomous driving—so what can YOU think of AI doing next?
Now, we want to know what YOU think! Are self-driving cars the future, or should we pump the brakes on this whole idea? Drop us a comment and let’s get that conversation revving!
IntelliPrompt curated this article: Read the full story at the original source by clicking here