Dr Roshy John
Practice Head, Tata Consultancy Services
When did you start working on driverless car?
I started working on this project more than five years ago, as a hobby, after having a near death experience in a taxi cab with an overtired and sleepy driver.
Is this project funded by any research institute or industry?
The whole project is funded by my savings and the money I pooled in from my patents. I took up this project more as a social commitment as I aspire to do my part by coming up with a technology to reduce road accidents.
What were the guidelines in selecting the team of engineers to work with you on this driverless car project?
The concept of autonomous vehicles is largely based on mobile robots, where we use software algorithms on digital filters, probability theory and advanced mathematics. Apart from these, one should have deep knowledge in advanced electronics. When I started my career, one of the most important things I did was to groom up team members to gain experience in the above said domains.
A robot typically has 20% of hardware and 80% of software, irrespective of its mechanical complexity. I have years of working experience in developing robots for numerous industries starting from factory automation to consumer electronics robots. My PhD was in the same domain on multi-mobile robot interaction. This helped me in grooming up my team members to reach a competency on some of the niche areas in Robotics.
On top of these, the entire crew was passionate enough to develop a prototype at least to this scale.
What were the hurdles faced during the years of leading this project?
The major problems we faced are:
- The rapid rate of battery consumption of the driverless car computer
- Actuator latency and backlash for the pedal robots
- Hardware failure when tested for extreme test cases
- Inability of radars and LIDAR identifying certain dynamic obstacles, viz. other moving vehicles or pedestrians
- Developing the AI algorithms on decision making to react during an emergency scenario
- Finding the probability of pedestrians, two wheelers and autorikshaws taking unpredictable lane changes
Having successfully demonstrated the driverless car, what is the next step?
We have made a small leap towards driverless car. There are numerous issues which have to be addressed, such as making cheaper and energy efficient hardware and smarter software which could handle Indian road scenarios. If the driverless car technology can successfully work in Indian roads, it will work in any part of the world!
What according to you is the future of driverless car or say assisted driving?
There would be tremendous development in this domain since the price of electronic components and sensors are coming down drastically. Almost all the car manufacturers are going to adopt ADAS and limited self driving as a feature within the next 5 years. Apart from these, the infrastructure to support these technologies is getting developed in an exponential rate. For instance, faster mobile networks, more precise GPS, etc.
In our regional context of India, what is the utility you see for driverless car or semi-autonomous cars/assisted driving?
There are tremendous opportunities in India and there are two ways to look at it. One is for catering the local market here the growth is exponential where every manufacturer is releasing newer versions/variants of vehicles every year. The second aspect is bringing more jobs to India. Throughout my career, I have focused on bringing these types of high-tech jobs to India. We can develop all these advanced technologies for supporting the foreign manufacturers or OEM suppliers to do this at a fraction of the cost.
Would you like to comment on the ecosystem in India for such projects?
We need to scale up our capabilities, resources and investments to execute projects like these which have a lot of potential in the international market. If my small team can contribute this much by making a driverless car, imagine what the bigger research groups in India are capable of!