|
Listen to this story
|
The next wave of AI is likely to be Physical AI or Embodied AI. Today, a number of startups are focused on building this technology, including the California-based Vayu Robotics, which is strategically accelerating autonomy.
Safety, Cost, and Autonomy
“If autonomy is ever to make it into the world and gain trust and credibility among the people, then it has to start at a very safe spot. So the biggest knob that we have to dial up is the safety, [by] reducing the mass and speed of the robot,” said Mahesh Krishnamurthi, the co-founder and chief product officer of Vayu Robotics, in an exclusive interaction with AIM.
“The kinetic energy of our robot is almost 1000% lower than the kinetic energy of a truck going on the road. So, by reducing the kinetic energy and the momentum of the robot, we inherently get very high safety,” he said.
Addressing safety is one aspect, but another important focus of the startup is developing new sensors that are both low-cost and high-quality.
“So, we thought, maybe we could take a bet on this sensor and say that maybe in a few years, we would be able to build a sensor that is super low-cost, but very high in quality. That would bridge the gap between a very expensive LiDAR system, which is very high quality, and an inexpensive camera system that is not as safe as a LiDAR.
“We tried to solve this problem and came up with the technological breakthrough,” said Krishnamurthi.
Vayu Robotics is focusing on developing an intelligent drive agent that requires less capital to build by using simulated data for training, unlike Tesla’s approach, which relies on real-world data.
Krishnamurthi believes that in recent years, simulators have advanced to the point where their output is nearly indistinguishable from reality, allowing them to create driving behaviours in simulations and seamlessly apply them into the real world.
“So that’s another technology breakthrough that has enabled us to build a product like a delivery robot driving like a bicyclist for less than $5,000,” he added.
The startup has developed six specialised simulators, each tailored to specific environments. These include one for indoor environments, one for outdoor settings, another for bike lanes, and yet another for the highways.
Additionally, they have a dedicated simulator purely for validation purposes, which is kept separately from the training process.
Vayu and Strength
Inspired by the Sanskrit word ‘vayu’, which means intelligence behind all motion and energy, Vayu Robotics is building to enable autonomous mobile robots to move through the world. Not just that, the founding team comprises members with decades of experience in the autonomous segments.
Krishnamurthi comes with a deep-tech background, beginning at Intel Labs in 2008 and moving on to Apple, where he worked on autonomous systems. Specialising in optoelectronics and LiDAR technology, he later joined Lyft.
He eventually co-founded Vayu Robotics in 2022 with his friends Anand Gopalan, who was the CEO of Velodyne Lidar, and Nitish Srivastava, a former colleague at Apple.
“Since we have collectively worked on this problem for about 25 years, we kind of had an insight on what works, what doesn’t, and what just might,” said Krishnamurthi.
The startup closed its seed financing round in October last year with backing from some of the biggest VC players, including Khosla Ventures, Lockheed Martin Ventures, ReMY Investors and others.
“Khosla himself is one of the big backers of this vision. He loves the idea and he is the best backer we could have as a startup. A true visionary of what the world needs to look like. We are fortunate to have him,” said Krishnamurthi.
Over the next few years, the startup is focusing on commercialising and scaling their technology, with a contract to deploy up to 2,500 robots across multiple US cities in phases. It will be a scaled ramp-up, with set milestones.
The startup will also focus on building sensors as a product on its own. “We will also try to have some ancillary applications for just the sensor,” he said.
While a number of robotic startups are emerging in this space, Krishnamurthi doesn’t perceive it as competition, rather a combined effort to solve a common problem.
“I do feel like it’s a good time for the industry and people in general because more people working on a problem increases the probability of it getting solved,” he said.
Without revealing the specifics, Krishnamurthi mentioned that their primary customer is a multi-billion-dollar e-commerce company, and another major fan company has expressed interest in their sensing product, making them another key customer.
