At CES 2015 NVIDIA (and Audi) showcased a computer , Drive PX – based on the TEGRA K1 chip, designed to be embedded in a car to provide an understanding of the environment.
NVIDIA is possibly the most advanced company in image processing chips, we find its chips in many computers to manage the graphic processing, in many game stations and also in many cars for the navigation system.
At CES they have presented a system that can process up to 12 video feeds from cameras embedded in a car and that uses deep learning architecture – software to recognise objects. It can tell that there is a garbage bin on the sidewalk and that that bumps just behind it is actually a kid to watch out.
The Drive PX can support a variety of sensors and easily provide 3D graphic rendering for display on the dashboard. It can be monitored remotely and its software can be updated to fix problems and support new features (something that so far only Tesla seems to be be doing, with the other car manufacturers updating the car computers software once you take your car for maintenance to their shop).
Drive PX marks the graduation of deep learning from being an advanced technology to be used in specialised labs and applications to become a mass market technology. Deep learning mimic in a way the architectures and computations occurring in our brains …
We’ve got to watch out: our cars are getting brains that for the better or the worse will be competing with ours. I just hope they will suffer less distractions that I am experiencing …