Dr Chris Jones Interview: iRobot Research Chief On Why Building The Ultimate Robot Isn't The Ultimate Goal

Posted: Updated:
Print Article
ROBOT
Alamy/iRobot/HuffPost

Dr Chris Jones is Director for Research Advancement and Senior Principal Investigator for iRobot Corporation.

Yes, he has a job title right out of a Isaac Asimov short story. And, if life was as narratively just as fiction, he would be either a megalomaniac building giant robot spiders for his own twisted delight, or a doomed visionary destined for ruin at the hands of an automaton militia.

He is neither. Instead he is a thoughtful, practical R&D scientist, who just happens to have been building commercially successful robots for more than a decade.

Yes, that's a thing now. And in a way, it's even cooler than fiction.

irobot roomba 780 1

Above: The Roomba 780 Vacuum Cleaner

Robotics is an area of tech that like voice control, 3D video and virtual reality, still seems like a futuristic dream - even though it has existed in practical forms for many years. Hollywood still busies itself by inventing robot fairytales for blockbuster movie, but you can actually go to Curry's today and buy a robot for a couple of hundred quid to do your housework -- and then drive it home in a car that was probably built by one.

If you do, there's a good chance that robot will have been made by iRobot.

For more than a decade iRobot has sold its highly successful Roomba series of robotic vacuum cleaners, to the point where total sales now number more than 8 million units. The newest Roombas have come on a long way in ten years. The latest models are able to clean floors more effectively than ever, as well as being quieter and more intelligent than before. iRobot has also found success building land and sea robots for the military and first responders, which have recently been used in the wreckage of the Fukushima nuclear plant in Japan. Its underwater robots were also used in the Gulf of Mexico after the BP oil spill.

Still, Jones admits that there is still a gulf between the robot we all know should exist, and the robots that do - and actually have a reason to.

"It's just not all that productive if you're going to sit in the lab and try to build the ultimate robot," Dr Jones told the Huffington Post UK. "But to take what you've got now and understand how you might apply it, in a valuable practical way, while in parallel continuing to develop more advanced capabilities that in future iterations you could roll out, that's useful.

"Perhaps to satisfy the dreams of what a robot should be in a lot of people's minds, that's a complex problem that is going to take time," he said.

"But I would point out for example that a robot doesn't have to look like a person to be useful. Solving a problem where you have a robot with two legs that has to balance, and walk around and all that, is difficult. But at the same time people like iRobot have things like the Roomba, which operates on its own and doesn't look anything like a person but is very practical.

You don't have to satisfy the ultimate vision of a robot that looks, acts and smells like a person, you need to focus on what is the task you want it to do and what is the most practical solution."

Which is not to say that robotics in general - or companies like iRobot in particular - are not making progress. There are a number of strange and interesting areas where technology is constantly pushing back the boundaries of robotics and finding new solutions.

For instance, if a robot is to act intelligently in the real world it has to be able to look around it and know what it's seeing. It used to be thought it was possible to just show a robot a bunch of pictures of various objects, and have it learn to recognise them by blunt force. It turns out that isn't good enough. In a moving environment, with a standard camera and limited power and processing, a robot has to be able to recognise objects on its own and decide what they are for itself. And like it or not, that isn't easy.

"Having the robot be able to sense, understand and really respond to a variety of situations beyond what the products can do today," Jones said.

The best research iRobot has done is getting close to that practical reality, but it's still a long, long way off from anything like the perceptive ability of a human.

"Right now we have a system on the research side ...[that can] recognise about 10 different object classes. For example an object class might be 'car', 'horse', 'person' - and the system is trained to recognise that class. Beforehand you'd train the subject by showing the system a large number of pictures of the object... but there are other additional challenges you have to address."

Another area being looked at inside iRobot's labs right now is how to make a robot act in a way that is not just useful, but, well, pleasant

"When you think about it psychologically - how can robots be the most adaptive operating around people, having them act in ways that people find to be natural and accepted. That becomes even more difficult."

Part of that effort is building robots in different ways. Jones says iRobot is working actively to build robot arms and limbs out of inflatable materials rather than heavy metal, so that in an industrial environment they can work alongside humans more safely. Other companies are looking at this too.

When it comes to iRobot's work with military and first response robots, it's surprising to learn that there is actually a lot of crossover between the two arenas.

Did you know, for instance, that the algorithm which tells a robot vacuum cleaner where to check for dust derives from one which swept fields? For mines?

irobot 510 packbot 4

Above: the iRobot 510 PackBot

"That approach directly inspired the coverage software that the Roomba uses to vacuum your floor," Jones says.

"There's always fun stuff to do! I've been working in robotics R&D for over 15 years and it continues to be interesting. The ultimate vision is a ways out. There is a lot of practical stuff we can do along the way but there is always more to do."

But what about that ultimate robot he mentioned earlier. Surely Jones has a picture of what that might be?

I ask him and he takes a long, long pause. You can almost seethe shining, brilliant masterpiece in his head before it vanishes in a puff of reality.

"That's a hard question to answer," he said. "I'm looking forward to seeing robots and building technologies that are important for the long term. You know, making the robots smarter - that's something that is a hard problem and we're going to continue to work on it."