Being able to stay in your own lane is basically driving 101. Not too hard right? Most autonomous vehicles accomplish this fundamental driving skill by using sensors and cameras to detect the white and yellow lines.
But what if there are no lines? Rural roads don't even have pavement, let alone road markings. And in construction sites, drivers must ignore the lanes and follow the cones. How will driverless cars navigate through these areas?
Nvidia has the answer. The tech company developed a self-driving system that knows how to drive in these situations. The system runs in a prototype car called NVIDIA BB8, which can drive on roads with or without lane markings, steer around obstacles in the road, and guide itself through construction sites.
The secret is machine learning, which is a process where computers can use artificial intelligence to learn how to complete actions that they were not explicitly programmed to do. But unlike the new Audi prototype that uses trial and error to park itself, NVIDIA can literally watch and learn. A human driver will demonstrate a driving situation 20 times (at different times in the day) and the car will observe the process and eventually be able to duplicate it.
It can even apply what it learns from one driving situation to another. Watch this amazing demonstration video to see what else NVIDIA BB8 can do.
More car tech stories: