)
After retiring as co-CEO of Advanced Navigation, a deep tech company spawned by his university research that would go on to raise $134 million as it pushed the boundaries of science and secured global clients, Xavier Orr has returned to the world of research with a new venture.
Having co-founded Advanced Navigation in 2011, developing AI-backed applications for inertial navigation to service industries including space, defence, mining and sub-sea surveying, Orr left the company in July last year for family reasons.
But it didn't take him long to lay the groundwork of his next business, taking advantage of the advent of improved affordability and quality of humanoid robots from China.
Teaming up with two researchers he'd met at AI and robotics conferences, Orr pre-seeded $1 million for his new company Anitron, which began as a passion project in September to "test the waters" experimenting with AI applications for Unitree G1 robots from Hangzhou-based Unitree Robotics.
It was a full circle moment for the 2022 Sydney Young Entrepreneur of the Year, who in university in 2007 had tinkered with the Honda ASIMO robot.
"It was pretty impressive for the time. It could walk and do things like that, but the hardware just wasn't there - it didn't have the processing power, the AI wasn't there. I remember you could get the hands to manipulate objects, but then it would lose the ability to walk," he says.
"18 years on things have changed quite significantly, and now robot hardware on the market is coming out of China that’s really impressive, and it's in mass production."
Inspired by breakthroughs in "nature-inspired AI", including research that mapped the fruit fly brain, Orr's vision for Anitron is to develop and refine lightweight, adaptive general intelligence for humanoid robots.
As the robot hardware works on an open development platform, the company creates its intellectual property for the robot brain as software so that it can complete complex tasks. Anitron is gearing up for a beta trial to deploy eight robots across a variety of industries, from aged and disability care to housekeeping to construction, and see which works best.
Anitron has already achieved what is considered to be an important benchmark not only for functional adults but also robots - being able to fold a shirt.
"You’ve seen the big guys like Boston Dynamics and Tesla and Figure AI trying that challenge; folding a shirt may sound like a simple task, but it's actually quite complicated because the fabric can fall in an unpredictable way. You can't fake that test," Orr says.
"Some tests with robotic manipulation you can fake it with a pre-programmed routine, but you can’t do that with a shirt. That's why the folding the shirt demo is a good one.
"We’re able to get that done in about 8.5 seconds, and then we’re doing a house search for an unknown object or a specified object. It can open cupboards and remove items to search for this object."
Anitron is now working towards a full laundry demonstration where the robot will be able to search a house for dirty washing, take that dirty washing from multiple rooms to a laundry, wash it, then take it out, put it in the dryer, fold it, and return it to the room from which it came.
"That’s our next milestone, and that’s the point at which we’ll start a beta trial where we start deploying into industry," he says.
"We’ll deploy eight robots, and then monitor them very closely to see how they actually perform in the real world, in real jobs.
"Of those eight beta trials, we'll see which one goes the best, and then the idea will be that we can start leasing robots into that sector to do the work."
He says the company will target applications within industries or jobs where there are currently labour shortages, such as electricians, plumbing, health care, food preparation, agricultural farmhands, equipment maintenance, retail and manufacturing.
"At the top level of our AI brain is an LLM (large language model) which is kind of like ChatGPT, so you can talk to it and you can give feedback if a task is done improperly. It can improve on the task the next time," Orr says.
Orr's own funds have been able to support the business to date, but this month he expects to secure more liquidity through a seed round that will be used to hire more AI researchers and robotics engineers, as well as support staff to run the beta trial. The scientist and co-founder says there's no need yet to look overseas for funding.
"We’ve probably got enough demand in Sydney. We'll just stick with Sydney for now – this industry is a bit of a hot one, so there hasn’t been a shortage of interest," he says.
"It's early days for robots. While everyone would love to have a general-purpose robot that can do everything, I think we're a bit more realistic in that there's things that the robot is going to be good at, and there's things that it's not going to be so good at at this time.
"What we want to do is identify what it's really good at and those are the markets that we're going to try to service straight away."
)
