Startup Saturday: Steering in the smart driverless future
DriveBuddyAI aims to ensure road safety by implementing driverless cars at city-level.
Entrepreneurship is all about risk. Starting a venture even in old and familiar industries is ridden with risk. Be it an e-commerce business, a food app, energy saving or anything else, risk is a constant companion of an entrepreneur. But when a start up ventures into new and untested territory, the risk multiplies. This is exactly what DriveBuddyAI, an autonomous vehicle technology company, had to go through.
DriveBuddy aims to eventually setup systems for driverless cars. All is very well, except that the current scenario is such that driverless cars in India are not allowed. So what is it that drove the three young men, Nisarg Pandya, Harish Khadsalia and Saurav Wani, to venture into a business that seems to have almost no opportunity? We investigate.
The trio was working as software engineers in the IT industry when Nisarg’s curiosity was aroused by news about autonomous driving across the world. “I thought the world is changing gears and here, in India, our driving conditions are still so pathetic. I decided to do something about it,” said Nisarg. With this determination, he decided to start a company that would devote its energy towards developing driverless cars.
“The concept of driverless cars has two components. One is where you control the engine and the second is where you get live inputs from the road. Indian roads are very chaotic, which makes the environment for a machine to drive itself very difficult,” added Nisarg.
He felt that existing technologies were sufficient to control the engine, but what was most vital in autonomous driving is the input from the roads. Nisarg and Harish quit their jobs and setup DriveBuddyAI in 2016. “For a year and a half, we worked on developing systems that would help us collect data, which meant extensive hardware and software inputs,” said Nisarg.
DriveBuddyAI now has developed a gadget that can be fit inside a vehicle to give live feeds regarding the vehicle and the road that it is on. “Essentially, you need to know three things. One is where you are. The GPS can give you details regarding latitude and longitude, but that is not enough. You also need to know what is around you. So we have a camera that is fit onto the dashboard that gives a 160 degree wide angle and 120 degree vertical view. This will help us know what the hurdles on the road are and will inform us whether there is a car, a human or a tree in front of the car,” said Nisarg.
According to him, the third thing is to be able to calculate the distance and the speed. “If the camera sees a dog on the road, we need to know how far it is from the car and the speed at which you are driving to be able to calculate what and when you need to take action. It will determine how you fast or slow you need should slow down. For this, we have two types of radars. One is the regular radar which will point out what object is on the street. The other is a laser radar called Lidar, which can calculate distances. With these inputs, it becomes easy to take action, irrespective of whether it is being controlled remotely or by a driver inside the car,” added Nisarg.
Artificial Intelligence cannot work without such inputs, said Nisarg. “Our roads are very chaotic. In some cities, the signals are mid air at the chowk and in some other cities, it is on the left and right corners. There are too many variables. There may or may not be lane markers. For a vehicle to drive itself, it needs to know everything. For driverless cars, precision is of paramount importance. So the GPS will tell you where you are. But if you were to go into a tunnel, you will lose the signal and would be lost on our radar,” added Nisarg.
“To combat this, we have an Inertial Measurement Unit (IMU). So we have the previous location (before you enter a tunnel) and IMU will know what angle you are moving, what degree you took the turn at before getting into the tunnel. The inertial measurement unit will calculate the turn and help localise your car. It’s similar to mobile phone games where you do not press any buttons, but move your handset to give feedback,” said Nisarg.
While DriveBuddyAI has developed the systems required to get real time info on vehicles, it is still far away from autonomous driving. “Our transport minister has said that there will be no driverless cars in India as it would impact jobs, but we see huge opportunity abroad. Across the world, driverless cars are gaining momentum only in the US, China and Germany. I think it will only be a matter of time before India too gets into the act. In the US, three companies, Uber, Google and Tesla, have permissions to test drive their vehicles on public roads,” added Nisarg.
One such test drive ended in a disaster when a female cyclist was run down by an Uber driverless car. So how is DriveBuddyAI planning to raise revenue?
“Just three months ago, we finalised our gadget that can be retro fitted onto cars or buses. Since real-time data is critical to autonomous driving, we plan to first collect real time data about our roads. We are targeting fleet owners who have commercial vehicles and need to know more about how their drivers are driving their vehicles. So we have installed it for Maksons Auto in Baner, Transline Auto in Lohegaon and an insurance company on a pilot basis. What they get from us is real time data about their cars. We will have a subscription model where we aim to charge them a monthly fee of Rs 1,000 to Rs 1,500 per car and get data on a daily, weekly or monthly basis. One of them wanted a daily update as their cars are driven by different drivers daily. So far, we have made Rs 30,000 every month and we are aiming to get more customers.”
Interestingly, this was a model that was explored by Tata Elxi in Bengaluru, but was soon abandoned on account of regulatory issues. If the Tatas see no future in the business of collecting data, what does DriveBuddyAI see?
“Making a product that gives data to a commercial fleet must be too small a thing for Tata. But we see huge potential. Once we have enough data from many vehicles, we can use it for many purposes besides autonomous driving. This data can be useful to PMC, who can then know which roads are prone to accidents; to insurance companies who can get real time images of an accident to know exactly how it happened; and to the Regional Transport Office to determine a driver’s capability. Mobili, an Israeli company, has tied up with the Bengaluru municipal corporation to helps its drivers avoid collision. We also propose to do the same thing. There is nothing like this in India. We have dongles put in the engine that can give details about it and GPS that tracks a vehicle, but nothing that helps a driver drive better and helps him avoid a collision.”
Now that they have developed the beta version and deployed it onto three fleets, they are looking at raising Rs 3 crores from venture capitalists (VC). “We are monetising our data collection so as to fund our autonomous vehicle research. But we will need more funding and will approach VCs now that we have proof of concept, data collection and collision avoidance.”
While autonomous driving may still be a far-fetched idea, DriveBuddyAI is trying to make driving safe in India, which in itself is a huge goal.
Manmeet Rajpal, owner, Translines Auto
I have just put the device in one of my cars. I have a fleet of 50. I thought it is a good idea to keep a check on how my drivers are driving and whether they are safe. It also helps me keep a track on how often they overtake, whether they are following rules and so on. We already have a GPS that lets us know where the car is, but this will let us know how safe our driver is. It’s a good thing, but I think that the cost may be an issue. For a fleet owner, spending Rs 1,000 or more per car will be too expensive. Also, in addition to that, I will need to employ one person to manage the data that will be generated. So unless it is really cheap, it won’t work for me.
Aditya Makharia, owner, Maksons Auto
I got attracted to the gadget because it will gives me details about how my drivers are driving. That’s really good. But the cost will be an issue. I think a fair price would be Rs 300 per car per month. Unless a customer wants it specifically for a car, the cost cannot be justified for a fleet owner.
Ajay Agarwal, entrepreneur, mentor and investor:
I feel that despite whatever regulations we have now, driverless cars are here to stay. It is a technology that no one can stop. This is a technology driven vehicle. Cars are being manufactured by auto companies. So any startup that gets into this space needs to be tech ninjas. The ones who will succeed will have the technological edge. India is better placed than other countries in this space simply because we offer challenges that no other country does. Our roads are the most challenging ones in the world and our engineers work under tremendous pressure. I am pretty gung ho about the technology and Pune, with so many auto companies, will be a good space to attract buyers.