HYDERABAD: In the age of self-driven cars being made a reality by companies like Google and Tesla, an Indian research team from the International Institute of Information Technology Hyderabad (IIIT-H) is attempting the desi version of a driverless car. The aim is to make an intelligent car navigate the environment and control itself accordingly especially on Indian roads that has an eclectic mix of lane-cutters, rash drivers and traffic violators.
The team of engineering students, their professor K Madhava Krishna from the computer science department at IIIT-H along with Uurmi Systems, a company that deals with robotics-related projects, is crafting a prototype driverless car (slated for completion next year).
“Safety is the biggest concern. We feel like we could bring down the rate of accidents caused by human error by making a car self-sufficient,” said Akhil Nagariya, one of the core members of the project and an M Sc Computer Science student at the institute.
Two model hatchback cars, one petrol-operated and one electric — a Maruti Zen and Mahindra Reva — are the models the team has been working on. “By making this car a possibility, human errors like fatigue, poor vision, drunk driving, rash driving and others will be cut down, slashing mortality rates on roads,” Nagariya explained.
He said that the 3D laser detection technology used by large overseas companies can run anywhere upto `40-80 lakhs, making it unaffordable for all. So going in for cheaper alternatives, the team is looking for ways to minimise manufacturing cost. The current car models being designed in the lab will have three sets of key features, the primary ones being two stereo cameras to pick up on colour and depth of objects around the car, serving as the eyes. The secondary feature will be the IMU (Inertial Measurement Unit), which comprises gyroscopes – commonly found in gaming smartphones — that aid in motion sensing and accelerometers.
Both understand movements and objects around the vehicle. The GPS system embedded in the car will help in navigating, which will be enhanced due to a pictorial and sensory understanding of its own location.
“A combination of detectors are used in the model, so even the 5-10 metre error that is possible on a GPS system is largely cut down in the prototype,” said Akhil.
The initial version of the car, funded by the institute and Uurmi Solutions, will be able to perform tasks like driving, navigation, understanding traffic and traffic signs.
The car can also change routes and merge with intersections.
Unlike the smoother roads of London or California where other versions of automated cars have already run trials, the challenge is vast for these youngsters who have a vision of making it possible for our terrain. “Instead of taking the cloud-based route by storing data for navigation, we are taking a real-time data processing approach,” he added.
“Most of us in the team have robotics as our electives, and our institute’s robotics lab has been very useful in testing out our models on smaller four-legged ‘Husky’ robots that move like a car.”
The team from IIIT-H, along with others from around the country, took part in a contest for institute-level students devising driverless technology for cars in Chennai. It was conducted by Mahindra & Mahindra Automobiles.
Here’s How It Works