Full-Stack Spatial AI: Taking Commercial Robotics into the Mainstream
Commercial robotics is on the cusp of going mainstream, but for that to happen companies need to stop having to reinvent the wheel. Lucy Ingham speaks to Owen Nicholson, CEO and co-founder of SLAMcore, about the game-changing potential of full-stack spatial AI
It’s hard to over-state the potential of commercial robotics. Over the next five years the field, which covers everything from drones and medical robots to delivery and agriculture robots is set to rise dramatically, with a projected compound annual growth rate of 33%, according to Mordor Intelligence.
Those working in the field are no less optimistic. Marc Raibert, CEO and founder of Boston Dynamics said in 2017 that “robotics will be bigger than the internet”.
Owen Nicholson, CEO and co-founder of SLAMcore agrees.
“We really do see the robotics industry as being something which really will revolutionise the way we live our lives in a really positive way,” he says.
“There was $2bn of investments made in Q1 of this year in robotics companies. So this market is already here, but it is set to grow exponentially over the next 10 years.”
All of these robots – no matter what their application – have one thing in common: the need for spatial AI, also known as simultaneous localisation and mapping (SLAM).
“Spatial AI is the ability for a robot to translate the sensor data it has onboard into an accurate representation of the space around it,” explains Nicholson.
“'The success of the entire robotics industry is reliant on the ability for these questions to be answered accurately, but really, really importantly, because ultimately we are a business, in an affordable way.'”
He argues that this covers three areas: its position, which is generally handled by a GPS chip; its awareness of the world around it and its ability to identify the objects in its vicinity and categorise them appropriately.
“Those are the three fundamental spatial questions that any autonomous machine needs to answer, and actually when you think about it, when a system fails, if it crashes or if it gets lost, it's actually a failure of the spatial AI, the ability for the robot to answer those three questions,” he says.
“So it's ultimately a discrepancy between what the robot thinks is its spatial reality and what the actual reality is.
“The success of the entire robotics industry is reliant on the ability for these questions to be answered accurately, but really, really importantly, because ultimately we are a business, in an affordable way.”
Spatial AI and the need for a robotics industry supply chain
All forms of robotics require some form of spatial AI system, whether it’s a basic technology that simply handles a robot’s location, or a solution that covers all three aspects, which Nicholson describes as full-stack spatial AI.
However at present, companies are faced with the task of building their own in-house solution. And this poses a significant barrier to the robotics industry meeting its true potential.
“The problem is that the robotics industry is still very new, it's kind of emerging, and it's almost stuck in a rut where a lot of the companies are actually building spatial AI systems themselves in very isolated environments, silos really, and that's the nature of an industry that is still maturing, because ultimately the supply chains haven't been defined yet,” explains Nicholson.
“As an industry gets more mature companies will emerge and own specific parts of the stack, but we're still entering that phase now. We're kind of where the automotive industry was back in the early 1900s, where the only people who could make a car were people who did everything from scratch.”
“We're kind of where the automotive industry was back in the early 1900s, where the only people who could make a car were people who did everything from scratch.”
While companies are developing some impressive robotics solutions despite this limitation, the lack of a robust supply chain is hitting in one particular area: cost.
“To build a robot there's lots of things you need to do, and spatial AI is only one part of the system, but to do it well is phenomenally complicated. There are literally only a handful of people in the world, and I'm talking dozens of people in the world, who are capable of building a full system at a level which is commercially viable. And they're nearly all locked up in the big tech companies, or they're so deep in academia that they wouldn't have any interest in working with a smaller robotics company,” he says.
“So what happens is these companies, they still produce high-quality solutions but they end up having to spend more money on hardware. They have to rely on more expensive sensors to address the problems; they're not able to have the internal expertise to do this in a way which allows you deliver the solution [affordably], while still having the high-performance and the high-quality.”
Making robotics affordable
By developing the robotics supply chain, the industry has the potential to become truly affordable, opening the door for robotics solutions that go far beyond high-cost, mission-critical applications and into every area of our lives.
And when it comes to spatial AI, SLAMcore hopes to be the company to make this happen.
“We want to have a huge, positive impact on the way we all live our lives, and the way we want to do that is by enabling robot companies, innovators to disrupt their markets and really catalyse this revolution of robotics that we believe is already here but is only going to get bigger,” says Nicholson.
Although only three years old, having spun out as a startup from Imperial College London, the company has already attracted $6m in funding and boasts a staff of 20, 14 of which are PhDs.
With early stage customers in the UK, the US, Japan, Taiwan and Singapore, SLAMcore plans to officially bring its full-stack spatial AI service to market in early 2020.
“We want to have a huge, positive impact on the way we all live our lives.”
And when it comes to specific industry verticals, the company is casting a very wide net.
“One of the most exciting things about this industry and also one of the real challenges is it’s so varied,” says Nicholson.
“Robotics can be applied to so many different industries: I'd argue nearly every single industry in the future robotics will touch in some way, and the challenge is every application is different, every environment is different and there's no such thing as a standard robot. I quite often liken it to nature: there's no such thing as a standard animal.”
Initially, however, SLAMcore will be focusing on commercial mobile robots, be it in retail, hospitality, healthcare or beyond.
“There's this very common technical requirement even though the applications are quite different, so it can be anything from supermarket stock checking robotics to service robots in hospitality or telepresence robots for helping people work remotely. But when you pick up, take off the lid and look under the hood of these robots, actually from a technical point of view it’s very similar,” he says.
“So that's our focus right now, these commercial mobile robotics, and particularly the companies that are looking to disrupt their industries, so bringing new products to market, or doing things in a cheaper, more affordable way, and we believe that's a really good market for us, because there are lots of companies in this space.”