Georgia Tech develop self-driving car capable of drifting on dirt tracks
Self-driving technology is being aggressively explored by automotive experts to one day eliminate manual driving completely. That dream is just in the infancy stages, although giants like Google and Tesla Motors are making impressive developments for the same. Autonomous cars are good, but don’t you think they are little boring? Always going by the book and not taking account for error by other motorists or unforeseen situations like bad weather conditions or a semi-truck losing control and hurling towards the self-driving car. In those situations an autonomous car would be more of a danger rather than safe commuter. To cover that situation, Georgia Tech have developed a technology which prepares autonomous vehicles for the rough rides as well.
Using a 1/5th scale model RC pick-up truck which is heavily modified, the researchers teach it how to autonomously drive on dirt track, and even drift while doing so. The RC truck dubbed AutoRally Robot is fitted with front-facing cameras, GPS sensor, quad-core computer, Nvidia graphics processor and 32GB RAM to do all the computing to stay on track.
After initially being semi-controlled by a human operator at speed of 18 miles per hour, the truck learned to go around the circular track on its own. Now, it goes at a speed of 62 miles per hour and keeps on learning how to navigate around with unpredictable dirt as its path.
Currently, Georgia Tech have loaded the software with 2,560 different trajectories for it to calculate the best way to take a corner, and come out without any hitches. Every second the software calculates all those trajectories 60 times, and decides the steering, accelerator and brake input. That said, it does falter on the odd occasion due to the driving conditions.