Home Defence, Training & Simulation Spatial AI: The all-important step in the robotics revolution

Spatial AI: The all-important step in the robotics revolution

There’s a small and incredibly important step we need to take before robots can join us in our day to day lives. 

SLAMcore is developing the algorithms needed to allow machines and robots to understand and navigate – really “see” – the spaces around them. We caught up with co-founder, Owen Nicholson, to discuss how good quality and commercially affordable AI can help us take the next step in the robotics revolution, and why Novatech is the hardware partner he chose to help bring his vision to life.

The beginnings of the SLAMcore vision

The story begins at Imperial College, where Owen was working together with the three other eventual SLAMcore founders – two academics and a PhD student. The way Owen tells it, what brought them all together was the concept of “getting out of the lab and into the real world.” Owen has spent his entire career working in innovation for everyone from Dyson to the UK government. But where his real passion lies is in robotics.

Slamcore - Novatech blog - The slamcore team

The Founding SLAMcore Team. Owen Nicholson 2nd on the right.

 “At Dyson, we were really interested in robotics and the next generation of vacuum cleaners. We were working with an Imperial professor who was a world leader in that space. Through that engagement, we decided to set up a top-class robotics research lab for him, and during that time I became fascinated by the implications and the practical applications of robotics.”

Owen was so enamoured by the professor’s vision for robotics that he decided to join him at the university lab to help set up the project. Here, they spent three years working together, gathering the two other equally single-minded, eventual company founders into their fold. At a certain point, they began to realise that there was a lot of interest in the type of work they were doing that extended far beyond intelligent vacuum cleaners. And that’s where the seeds of SLAMcore were sown.

He adds: “We wanted to bring vision as a product in its own right to the robotics community. That was the seed of the idea that grew into SLAMcore. It seemed like the right time too as even a few years ago you could see this trend coming through connected machines and the IoT revolution.

Check out our other AI and deep learning stories

SLAMcore: Position in action

The missing link – spatial awareness

As far as the new team were concerned, however, connected machines were limited by the fact they were not aware of their surroundings. As Owen explains: “They're just living in their own little world really. They might have some sensors installed for reading their specific immediate area, but they can't share that information in a meaningful way as far as where they are in space.”

That’s where his team’s solution comes into play - creating a spatial AI solution that runs on affordable hardware. Indeed, they’ve even been able to create a system that runs on a Raspberry Pi. This is what they see as the primary thing holding back the robotics revolution - the cost of entry.

At the most basic level, SLAMcore’s solution might allow a robot vacuum cleaner to recognise and keep away from the cats so that it doesn't scare them. Or, it could mean a warehouse with a system in place that means every asset (down to the very last spanner) has the ability to position itself and share its location. 

Since we founded the company four years ago, we've really spent the time building a quality solution and then making it in a way which is flexible, tunable, affordable and practical,” Owen explains. “Because one thing you can be absolutely certain of is that there is no one-size-fits-all solution for the robotics industry. There's no one animal in nature that thrives in every single environment, so why would there be one robot?

Slamcore - Novatech blog - Exploring real-world applications

Smart Robotics Make Distribution Centers Productive and Profitable. Image ©Honeywell.

Exploring real-world applications 

SLAMcore markets itself as a solution for all developers who develop robot systems. In that area, they see themselves as a small but incredibly important  “component piece to a much wider autonomous stack.” Their users are primarily robotics companies and those companies specialise in full autonomy.

Currently, they are focusing on industrial and logistical robots for use in the warehouse - robots that move pallets around, and forklifts that can be automated so that the humans can focus on the more dexterous parts of the job. The team is also looking at the low-cost robotic space, which is mainly based around things like automated vacuum cleaners and lawnmowers.

The output of our system is really just coordinates and a map. You can’t do anything with it on its own. You can almost think of it like we provide the GPS system and then it’s over to our customers to integrate that into a full product.” 

Every customer will have different needs, of course, but in most cases, the team is quite adamant that the sensors which make the biggest difference are those relating to vision. Concentrating on vision gives machines the ability to read and understand not only relative space, but also more intricate things like texture, colour and shape. 

Owen adds that using these sensors alongside neural networks and AI you can also “identify objects and segment them. So not only can you position the machine but you can also position the objects around it and it will know what those objects are.” 

All it requires is a stream of visual data that can then be converted into the information needed for the machine’s autonomy stack, which is accessed through an API or running on a local edge computing machine.

Making a difference in robotics 

For their AI and machine learning, SLAMcore needed access to the best hardware for building and AI algorithms, and that’s where Novatech came in.

Owen explains: “We have 30 engineers, all of whom have PhDs and all of whom require hardware that allows them to be as efficient as possible. We can't have them sitting around waiting for data to compile. So the better the hardware, the faster we can get our work done. That means we can stay ahead of the game. And that's just absolutely crucial, especially for a company that doesn't have the resources of a tech giant.” 

Ironically, for a company focused on robotics, Owen believes it’s the people that truly make the difference. 

“We have such talented people that can build these proprietary constructs and algorithms from the bottom-up. Because our team has that deep understanding, we can use AI throughout the entire spatial stack and that means it's so much more powerful than just a single element. It means we can create a system that recognises and understands everything around it.” 

Thanks to Owen for taking the time to discuss this fascinating step in the widespread adoption of robotics. 

We specialise in engineering custom, purpose-built IT hardware.

If you have a project you’d like to discuss, drop us a message using the form below, or call our team on 02392 322 500.


Posted in Defence, Training & Simulation

Author -

Published on 17 Feb 2021

Last updated on 17 Feb 2021

Recent posts