Autonomous or driverless car is no more going to remain a utopian dream and is now more likely to become a reality soon. However to become reality, the car must be able to answer three critical questions independently — Where is the Car? What’s around the car, and based on the responses to these two questions, what should the car do?
The third question can be replied to by a series of complex logarithms and software. Researchers at the University of Cambridge are seeking to answer the other two questions.
It seems that GPS is the best suited to respond to the first issue as to where am I? However, the GPS has been found waiting at times and has been inaccurate, slow and sometimes unreachable. The new system makes good by supplementing the GPS data with computer vision.
The new system excels the GPS in many aspects. Using a single color image, a computer can discern its exact location to the accuracy of a few meters and even a few degrees. It could also know which side of the building or structure it is facing, even if all the sides look similar.
The second system that has been christened SegNet again uses computer vision to answer the second question. The process analyzes any image that it has not analyzed before and quickly classifies and sets color codes for different objects into roads, street signs, structures, pedestrians, and more.
The biggest advantage of the system is that it uses an image clicked through a camera in your smartphone.
The device has already been tested at the King’s Parade in Cambridge, and lead researcher Prof. Roberto Cipolla said that it will be some time before the concept can be commercialized but with more accurate and precise technologies coming, a driverless car could become a reality in the future.