Deep Learning for Autonomous Cars

The current major paradigms for vision-based autonomous driving systems are: the mediated perception approach that parses the entire scene to make a driving decision, and the behavior reflex approach that directly maps an input image to a driving action through regression. A third paradigm Direct perception approach was proposed in [2], which maps the input image to a small number of key perception indicators that are necessary to drive safely. We use the power of transfer learning to test the direct perception approach by fine-tuning a standard AlexNet architecture pre-trained using ImageNet data. The data for training and testing are taken from the TORCS game. It is observed that the fine-tuned network performs similar to the network trained from scratch as provided by [2].