Dr. Nick Hawes is a Senior Lecturer in Intelligent Robotics at School of Computer Science, University of Birmingham. His research interests lie in application of artificial intelligence (AI) to create intelligent, autonomous robots that can work with or for humans.
He is also a passionate and energetic believer in public engagement about AI and Robotics, and science more generally.
In this interview for PlanetTechNews, Nick covers the trends in Robotics and the link with AI which could be most relevant for future development in the field; the topic is often of great interest for anyone following science news and technology news.
Q. How did you get interested in robotics and AI and what you consider as your particular expertise?
A. I’ve always been interested in technology and computers (my mum was a big scifi fan, which I think encouraged my interest). I chose to study AI as an undergraduate because I thought it sounded incredibly exciting, and I really wanted to do something different from the standard courses my peers were choosing. After doing a BSc and PhD in AI I wound up working as a postdoc on robotics projects as my PhD supervisor thought I’d be suitable. Since then I’ve never looked back.
In terms of expertise, I see myself as an integrator. I work to integrate academic ideas (you need to bring together separate fields in AI to make robots that actually work intelligently), software (I greatly enjoy taming the complexity of the large, heterogenous software systems necessary to make robots tick), and also people (as you need reasonably large, disparate teams to contribute to intelligent robots in an academic setting).
Q. What you see as the next important step in your career?
A. I think there are two important revolutions approaching. The first is the reintegration of AI and robotics. These were related topics at the birth of AI, but grew apart. Thankfully a collection of people are bringing them together again, and I think this will lead to smarter, more capable and robust robots able to work autonomously in the real world. The second revolution is driven by this, and is the appearance of these autonomous, intelligent robots in more and more industrial settings. For the next step in my career I’d like to play a role in both of these revolutions, ensuring they move in the right directions.
Q. Is there any area of business where robots are being underutilized?
A. I would say everywhere! But that’s mainly because the cost/benefit ratio is not right at the moment, even though the technology is almost ready for specific applications.
Q. Why has the original 60s style vision of ubiquitous humanoid robots not turned out the way many people expect?
A. Because the problems in building general purpose intelligence, coupled with a physical body, turned out to be much, much harder than anyone expected in the 60s. Computers, and thus robots, are great at doing clearly structured, predictable tasks, even if these tasks appear very intelligent to humans (chess is a great example of this). On the other hand, behaviors which seem easy for humans, and that we’d require in one of these predicted machines — such as seeing, walking and talking — turned out to be incredibly difficult to automate, partially due to the huge range of inputs and conditions such behaviors have to cope with. We also expect intelligence to degrade gracefully, that is to not completely fail when a problem can’t be addressed. Most AI and robotics work is still incredible brittle in this respect: when an unexpected situation is encountered, the results are often a long way from intelligent.
Q. How interconnected are developments in AI and robotics?
A. Historically they haven’t been, but this is starting to change.
Q. What are the key innovations in robotics you expect to see over the next 10 years and, as longer term predictions, for the next 30 years?
A. For the next 10 years I expect to see more robots performing clearly defined, specific tasks in a range of industries. I would expect these to be industries and tasks where the still significant cost of developing and deploying a robot to be worth it, and also where the robot’s performance is relatively safe (as the legislation for autonomous and learning machines in human environments is still to be addressed). I would also expect to see more and more teleoperated, or partially autonomous robots, working in dangerous situations such as disaster zones. The role of the human in their control is still essential as the technology is still to mature to complete reliability, and the user’s trust is hard to gain. It’s much harder to make predictions for 30 years, but I’d expect to see the trends I’ve already mentioned continue, with increased autonomy, intelligence and learning in our machines and software. I’d like to think much of travel, particularly in private cars, will have been automated by robotics technology by then.
Q. Would you consider robotics to be an accelerating technology, or is the future very much down to unpredictable breakthroughs?
A. At this point it’s an accelerating technology. AI is grounded in Computer Science, and I don’t expect huge unpredictable breakthroughs in the underlying logic and maths in Computer Science any time soon. As computers get faster, cheaper and smaller, and we get better at sharing and reusing our AI and robotics results and code, I think we will see steady improvement in the technology available. In terms of hardware, it would be hard to rule out unpredictable breakthroughs as I feel the space for innovation is much larger (innovation which software can then exploit). For example, the (unforeseen) appearance of the Microsoft Kinect has had a huge impact on what robots can sense, and therefore on what they can do.
To find more about Nick please see:
http://www.cs.bham.ac.uk/~nah/