PremiumSaurabh Chandra calls his three-year-old Ati Motors a ‘rogue startup’. First, it’s based in Bengaluru’s traditional Malleshwaram neighbourhood where uncles discussing cricket scores are a common sight—not a startup making autonomous vehicles (AVs).
Second, the AV space is a nascent one where tech giants and startups alike are hacking out the road ahead—so it’s definitely not a startup following a cookie-cutter model. Lastly, the three co-founders challenge stereotypes, spanning the spectrum of entrepreneur age and experience.
Fifty-five-year-old V Vinay is a former professor of computer science from Indian Institute of Science, while Saad Nasser is an 18-year-old prodigy who dropped out of Clarence High School when he was eight. The business fulcrum for the two tech wizards is Chandra, in his early forties, who earlier sold his IT services company to French multinational Publicis.
“After the exit, I wanted to start a school for gifted children because India is the only major country that doesn’t have such a programme," says Chandra. “That’s how I met Vinay who was running a weekend programme to mentor gifted children like Saad." The three realized they shared a love for toying with engineering problems, and they founded Ati Motors in February 2017.
MEET THE SHERPA
Three years down the line, they have an autonomous cargo vehicle called Sherpa commercially deployed in a textile factory in Chennai. But the factory is shut like most of India during the coronavirus lockdown.
So, the rogue startup has built a smaller sibling of Sherpa, which will be deployed in a Bengaluru hospital next month. AVs are already delivering medicines to patients in a Guangdong provincial hospital in China to minimize contact.
The Sherpa’s technology, which is similar to what’s being developed for driverless cars by the likes of Tesla and Google, is more versatile than warehouse or humanoid robots. Its eye-level LIDAR (Light Detection and Ranging) processes more visual cues than the worm-hole view of robots or humanoids which are fine-tuned to work in specific settings. This enables the Sherpa, which rolls on automobile wheels, to go around inside a factory as well as outside in the yard or road.
“We have taken a public autonomy approach and applied it to an industrial domain. Our resilience to changes in the environment is more. Our 3D LIDAR is sitting at a height where we see a lot more of the world than robots," explains Chandra.
A lockdown with empty roads seems like an ideal scenario to test driverless delivery of grocery and other essentials with autonomous vehicles like the Sherpa, which can carry a 150-kg load on its rear flatbed or haul 500kg in a trailer. Although such use cases are hard to pursue in India without regulatory support, grocery delivery in driverless vans is already happening in China and the US.
California issued a permit earlier this month to AV startup Nuro, founded by former Google engineers, to use public roads for grocery deliveries with driverless vans in the Bay Area. The startup is testing this in Mountain View before applying for a statewide permit for commercial deployment. On the other side of the world, Chinese startup UDI’s self-driving vans have been delivering groceries to communities in Zibo, a locked down area of Shandong province, for three months.
Chandra feels it’s inevitable that alternative systems will come into play with the awareness created by the pandemic. “What’s the business continuity plan for a city? Even if autonomous vehicles are not used regularly on public roads, they can be given ambulance-like status."
For now, however, Ati is focusing on deployment in factories and hospitals, because it doesn’t want to bite off more than it can chew as an early stage startup. It has mostly been bootstrapped, with support from angel investors.
The Sherpa has been tested on the IISc campus, where the startup has a collaboration with Robert Bosch Centre for Cyber-Physical Systems. For Professor Chiranjib Bhattacharyya from the Robert Bosch Centre, it was a chance to get hands-on with emerging tech. “It’s not easy to experiment with autonomous navigation on the ground. So that really excited me," he says.
The professor had been working with drones, but this was a full stack AV built from the ground up, including an electric drivetrain, electronic sensors, 3D LIDAR, on-board computing and the AI autonomy stack. It even has a suspension to withstand shocks on Indian roads and factory floors.
DIFFERENT FROM TESLA
The most exciting part for the professor was a chance to test algorithms for SLAM (simultaneous localization and navigation). This is a chicken-and-egg problem where the Sherpa has to map out new environments while simultaneously keeping track of its location using cues from the LIDAR.
Every new setting poses different challenges. For example, the Sherpa goes at a lower speed in a factory compared to a Waymo or Tesla driverless car on a spacious road. “Our problem statement is a little different from their problem statement," says Chandra. “They’re doing it at 80 miles an hour with their own 3D LIDAR whereas we’re trying to achieve precision at slower speeds in our context in the cheapest way possible."
Most of Sherpa’s SLAM algorithms come from Vinay, who earlier left IISc with three colleagues to build the Simputer at the turn of the millennium. This was a handheld device ahead of its time, with an accelerometer for the kinds of features we’ve now become accustomed to on smartphone screens. But India did not have a venture capital ecosystem back then to turn hardware projects such as these into global winners. Vinay moved on to launch other companies, including Strand Life Sciences for biotech and Jed-i, which mentors “gifted engineers."
Saad Nasser was a 15-year-old when he became a co-founder of Ati Motors, but he was already an expert at putting complex systems together. After dropping out of school, he built a full computer system with a mentor at Intel before Vinay took over his mentoring.
“Autonomy was very interesting because there’s a lot of integration work to be done across multiple disciplines," says Nasser. “It’s about integrating hardware systems with the software on top." For instance, everything had to be smaller for Ati’s new vehicle for a hospital. This also meant the LIDAR would be lower and algorithms had to take that into account to figure out navigation.
Sherpa’s rollout in a Chennai factory this year was a significant milestone, before coronavirus brought everything to an abrupt halt. However, the startup is getting enquiries from foreign markets, including California, and hopes to make a dent globally. Meanwhile, it will find its way in hospital corridors.
Sumit Chakraberty is a Consulting Editor with Mint. Write to him at chakraberty@gmail.com
