Is autonomous driving far from us? Perhaps not. If you look at the following road map published by Morgan Stanley Research, it is expected that in around 10-15 years we can safely let cars drive for us without much interference:
Great applications need powerful tools to realize them, and deep learning has been recognized as one such tool. Deep learning can potentially help in solving many problems that exist in the self-driving car system, for example:
- Lane detection
- Pedestrian detection
- Road sign recognition
- Traffic light detection
- Face detection/recognition
- Car detection
- Obstacle detection
- Environment recognition
- Human action recognition
- Blind-spot monitoring
And for all of these problems, it’s possible that we can leverage multiple data sources to gain better accuracy and speed. Basically, a self-driving car can be equipped with various...
