Learn by playing! Enhance your deep learning skills by making a self-driving car go around a race track as fast as possible.
Deep Learning for Autonomous Racing
This course provides hands-on experience with deep learning gained by solving problems related to autonomous vehicles. For participants, the aim is to train neural networks that control a car’s steering and throttle in a simulated environment.
Will a recurrent neural network built on sequences of segmented images win? Or will the simplest, fastest LeNet-like model be “the way to go” (pun intended)?
We conclude the course with a competition pitting the participants’ cars against each other. That’s a great opportunity to integrate your team and help them gain solid machine learning knowledge. May the best one win!
Team of machine learning engineers and software developers who would like to experiment, try out new ideas and to push deep learning techniques to their limits.
Engineering and mathematical maturity (Python / calculus / optimization). We use the Carla simulator (built with Unreal Engine 4) to drive around a racetrack. We first collect data, and then train a neural network to imitate our driving style. But no prior experience with racing games is required.