Driverless cars Soon to be a reality?

BENGALURU: Although India is nowhere near the conversation or innovation surrounding autonomous or driverless vehicles, two engineering students from a city college are trying to make some headway in this direction.

Sanatan Bhardwaj and Kiran Rawat are final year computer science engineering students who have designed a miniature prototype of a driverless car using the basic components and methodology that renowned companies such as Apple, Google and Uber are presently using to design their own.

The two students from MV College of Engineering, however, have brought in their own customisations in various aspects of the design. "The computational power of our driverless car system is not that powerful. We have therefore reduced the amount of data to be processed. In the present driverless cars, the computers processes data from the camera that takes not only information of the line of vision on the road but even other areas such as the sky using sensors. In our prototype, we have completely left out the sky from our camera's line of vision so as to reduce the load on computation," says Santan.

The team has also written their own computational algorithm and have conducted a few trials. "We have conducted trials on a small track of around 24 meters full of obstacles such as speed breakers and signals. I would say we are partially successful," adds Santan.

The prototype is 20 cms in length, 10-15 cm in height and a width of 15 cm. It has a four-wheel drive chassis used as a base. A Raspberry Pi (a kind of basic computer) used for graphic processing unit (GPU) and the CPU computations. A motor driver powers the prototype along with four batteries. The car uses a Pi camera and ultrasonic sensor to detect obstacles. The camera is mounted on a servo motor to make the head (camera) flexible to rotate.

The boys realise that much more work is needed on their model. "The work could be enhanced by improving the algorithm by adding machine learning to it. The present algorithm performs the operations on all the frames. It is accurate but its efficiency could be further enhanced if it starts learning by itself and avoids unnecessary calculations of the regions which are already known or familiar. Once the car starts travelling on the roads, it determines the obstacles (mainly static) on the way and notes their characteristic features," says a statement by the team.

While driverless cars are still a distant dream even in the West, Santan says that in India it is going to be even more difficult, considering the sheer density of cars on the roads.