Meet roboticist Nick Hawes, a lecturer in the School of Computer Science at the University of Birmingham. He is bringing his robot, Dora the Explorer, to our Robotville festival.
How did you become involved in robotic research?
When I was at school I didn’t really know what I wanted to do with my life, so I chose to study Artificial Intelligence (at the University of Birmingham) because I didn’t really believe that it was a real subject. I quickly fell in love with the idea of building intelligent systems, so stayed in Birmingham to do a PhD in AI for video games. It was only a few years later when I joined an ambitious project that wanted to use some of the ideas from my PhD to help build a cognitive robot that I moved away from virtual worlds and built things that ran in the real world. From then on I was hooked on robotic research.
How did you become involved in using robots for exploration?
We were interested in studying how robots could automatically extend their own knowledge about their worlds. We already had a robot that could build maps of its environment, but it had to be directed by a human using a joystick. Therefore it seemed like a natural extension to see if we could get the robot to take over from the human and guide the exploration itself.
What can Dora do?
The Dora robot can explore space to build up a map. It can choose where to explore based on how likely it is that a particular direction of exploration will produce interesting results (like previously unknown rooms). Dora can also search for objects, either in order to tell a human where they are, or to work out what kind of rooms are in the building (e.g. if it sees a kettle it will realise it is in a kitchen). Dora can also interact with humans in a limited way, by asking questions about the location of objects.
Do you think robots like Dora will become a household necessity?
I’m not sure they will become a necessity (unless we all become unable to care for ourselves!), but they will certainly become a luxury item in the future. It will be possible to order them from an online electronics store one day and have them arrive and start work the next. The big problem will then be training your robot to understand the layout of your home and the routines of your household. This is where we hope our work on exploration and curiosity will make people’s lives easier, as the robot will be motivated to learn things for itself, rather than wait to be taught everything.
Are the costs of transporting robots too prohibitive to send one to Mars? Submitted by Steve
The answer has to be no, as NASA has already sent two robots to Mars: Spirit and Opportunity. This has cost in excess of $900 million! One interesting question is not the cost of transport, but the difficulties of controlling robots once they get to Mars. The delay of a radio signal between Earth and Mars can be over half an hour, so this precludes any direct control by a human (just imagine playing a game where every command you send takes 30 minutes to have any effect!). Therefore these robots must be equipped with forms of Artificial Intelligence to allow them to safely make decisions for themselves with only limited human intervention.
Does the software that is tasked with being curious continuously increase its capacity to explore / make connections – is it getting more creative? Submitted by Charlie and Jake
I would say that it doesn’t really increase its capacity to explore (except beyond software that isn’t curious at all), but it does have a limitless appetite for exploration. Unless you artificially restrict its mobility, Dora will just keep trying to explore until its batteries run out or it gets stuck somewhere. A few years ago we were demonstrating Dora in a corridor that led to some toilets. Someone had left the door to the women’s toilets open, and before we knew it Dora had wandered off in there. Unfortunately, as our research team was all male at the time, we had to wait for someone female to come along to retrieve Dora!
Our view of robots is shaped by books and film – do you think this is helpful or misleading?
I think it’s mostly a good thing, as fiction (books, films and increasingly video games) motivate and inspire a large number of people to work in science and engineering to produce exciting and useful technologies such as robots. The main downside is that the reality of robotics research is a long way away from the images portrayed in fiction, so we risk disappointing people when we are unable to provide them with the robots that they imagine are possible. However, as you’ll see at Robotville, the robots that exist now are all really exciting in their own ways, and will only improve as more time and effort is dedicated to them.
Apple’s recent iphone update includes what some are claiming to be the first consumer robot. How relevant is this and do you think this idea of robots in our pockets will have a wider impact on robotic research and developments?
Whilst there have been consumer robots before Siri (including vacuum cleaners, lawn mowers and toys), the importance of Siri is that it demonstrates that it is possible to produce software that consumers can interact with about a range of tasks in a natural fashion (even if the results are not always perfect). Perhaps the biggest impact of the development of computers in our pockets will be the expectation that all our devices should share information to make our lives easier. For example my phone should know when I’m nearly home so that it can tell a future version of Dora to put the kettle on.
What are the ongoing cultural implications of your latest research?
Robots will change the way we live and work. The change will start from limited, well-defined tasks (such as robots that clean the floor and cars that park themselves) but will gradually become more noticeable across the whole of society. Some of the larger cultural questions we will ultimately need to face include whether we are happy to let robots care for our sick and elderly, and who is responsible when a robot that can learn and make decisions for itself causes some kind of problem.