Can you tell me about your background?
I’ve always been about robots, since the very start – I was always going to be making machines that move. So it’s no surprise at all that I’ve ended up here!
I was fortunate enough to get an undergraduate degree here at Oxford and that was wildly formative. What’s stayed with me was the mix of subjects that I got to work with – not just maths and software, but economics, politics, geography – I really valued that time and the way it changed my thinking.
A couple years later I went out to Australia to do my PhD. There was a moment when I was walking across the road, in my early days out in Sydney, and I turned to a PhD student who was finishing, and asked him what he was working on. He told me about a problem called “SLAM” – simultaneous localisation and mapping. It turned out to be a foundational element of modern robotics. By the time I’d crossed the road, I felt that that was the addiction – that was the problem:
Following my PhD, I took up a postdoc position at MIT where I did a lot of work on subsea vehicles. I started to think more deeply about the structure of the SLAM problem, and at the same time, my addiction for making it work on a real platform was fed and only growing. It’s quite a thing when you get an opportunity to write new software, stick it in a not-cheap vehicle, and press ‘dive’ – and know that it might not come back. That just fanned the flames!
When I came back after MIT to a faculty position here at Oxford as a departmental lecturer, I remember it very clearly: when I turned up, there was no room, no desk, no phone, no chair, nothing. But, I got the sense that there was also nothing in the way.
I left the subsea stuff – it was pretty clear that was going to be hard in Oxford, far from any ocean, and started building the group here (Mobile Robotics Group). I started to learn how I could work with industry. Two key things for me were that whatever we made, someone had to want it – it needed to be useful and solve a problem for people. The other key thing was persistence – how do you get these machines to work for a long time, not just because it’s a conference paper? So we started researching on duration, robustness, and all the time trying to put this software onto more and more ambitious pieces of hardware.
A big break was 2009 – 2010 where we won some big awards and started working with Nissan and BAE. We built a new lab, hired professional engineers, and started focusing on cars. I wasn’t interested in building an archipelago of ideas, I wanted a big mainland. But my colleague (and Oxbotica co-founder) Ingmar and I realised, we needed to go bigger, that we could really fix transport – and that’s why we started Oxbotica.
How do Oxbotica and ORI fit together? Why did you start Oxbotica?
The Oxford Robotics Institute is the place where we can afford to take long bets, and think about the problems that could be solvable in six or seven years’ time. For example, the stuff I was thinking about seven years ago is now running on the streets of London, in the GATEway project.
Oxbotica spun out from the university 2.5 years ago – I went to the university and knew we needed to do something different here – what we had was not just one idea, it was a whole load of ideas that come together. It’s the ‘where am I, what’s around me, what should I do’ – with many ways of solving those problems as the systems will encounter many different situations. There’s a tapestry of solutions you have to be able to bring together.
We had real substantial intellectual property payload, and the opportunity to invent on top of that – to take it beyond the stratosphere and out into orbit. Although we are now completely independent, just like any offspring, we still have a really close relationship with the university today.
What is special about Selenium?
Wow – that’s going to take hours! There are many things I like about it. It’s not one monolithic solution – it’s a tapestry – lots of dovetailing competencies that come together to provide flexibility, with a technical roadmap to where we’re going next. It’s something that means our customers can pick and choose from what they need most.
The fact that we didn’t take investment on day one means that our code has to be mean and lean because it’s got to be tradable. That means you can’t have a super computer in the boot – it’s got to work on a small computer. It can’t need a nuclear cooling system.
Of course – there are things I can’t share – the full technical roadmap and what’s coming next, all because of what we’ve done on the foundational technologies so far.
You’ve done a lot of public work in pedestrian environments – what about the roads?
There’s been a lot of public things – but of course that’s not the whole show! There’s a load of stuff that’s going on inside Oxbotica with clients on the road-going tech, and that will be coming out very soon. But roads versus pedestrian environments? It doesn’t make that much difference in the raw technology – you still need to answer the three fundamental questions of where am I, what’s around me, and where do I go next.
Sure – there are different things you need to do on roads, than if you’re going through a crowded pedestrian area. It comes back to the architectural question – and how we layer on to a system, for different applications. I’m sorry we can’t tell you everything about our various road-going applications, but stand by!
Do you need big computers to run Selenium?
No! In fact, two weeks ago, we just reduced the cost of the computers that we’re ordering because we need lesscomputing power. You really don’t want to open the boot, and us have to shout because of the roar of the cooling system! In short – Selenium runs on a single processing chip.
One of the things I really like is the culture we have here across the company – this delight in the leanness of what we write. We have established practices across the team to keep it that way, and part of our value proposition is the conversations we’re not having. There’s hundreds of years of thinking behind this – and a similar amount of rabbit holes we know not to go down!
What has been your most enjoyable moment in Oxbotica? What has been most challenging?
But that’s not because I want to show off about it, but because I feel great about it, and I want to share that – and that’s really important for the UK. But I’m always frustrated – that’s what drives me, that’s what engineers do. The drive is always about the next thing that doesn’t work – why does it not give into my will! One of the best moments is when you see the crew, having worked on things, sit in a car, and it drives itself. Those are real high moments.
Of course, there was the first cheque we got in. And since then, every single person that we hire gives me a buzz. You’ve made a job, because of some thinking that you and your colleagues have done. And the most extraordinary thing is that every time we do that, we get closer to more parents keeping their kids and more kids keeping their parents, because we can fix transport. So there’s a really nice, virtuous circle. It’s just great to be making stuff that has a moral imperative, as well as an economic one.
I genuinely believe currently, we move stuff in a stupid way, and that machines can help and they can do it better.
Tell me about this “spectrum” thing?
We have some clients who want the whole autonomy package, so that’s Selenium or Caesium. But then you have people who don’t really want an autonomous bus, they just want to know where the bus is, or they want it to be able to reverse without going over people – but still able to reverse over rocks, so you can’t do naive obstacle avoidance. And so Oxbotica offers a spectrum of competencies, that dovetail together into a bright white of Selenium. But, you can pick off colours from the colour cube, and you might want those as just standalone competencies, or you could combine them in different ways.
How does Oxbotica fit with the future world?
I’m not in a mood to hang around – I’m really not! So I really care about what the UK is doing to fix transport and moving stuff globally. I see us growing explosively soon – next month we’re moving into our new ‘Galactic Headquarters’ as I call it, where we have floors more space to keep growing. You go into the office and you really have a team who trust, like each other and really enjoy what they’re doing.
We’ll be providing software that helps more parents keep their kids and more kids keep their parents. We’re moving the needle on this, because of the team that we have, and it’s just awesome.
There’s hard work ahead for us, but bring it on!