Robust robots need to evolve – like babies or tadpoles – says roboticist

To build a really tough robot, let them be babies first, says roboticist Josh Bongard, He says the most robust robots are those that change their body forms while learning to walk.

A University of Vermont roboticist, Josh Bongard, has completed an experiment showing that – to build a really tough robot – you have to let robots be babies first. Or, to put it another way, you have to think about how tadpoles turn into frogs.

Josh Bongard

Bongard recently completed an experiment in which he created both simulated and actual robots that, like tadpoles becoming frogs, change their body forms while learning how to walk. And, over generations of robots, his simulated bots also evolved, spending less time in “infant” tadpole-like forms and more time in “adult” four-legged forms.

These evolving populations of robots were able to learn to walk more rapidly than ones with fixed body forms, Bongard said in a press release. And, in their final form, the changing robots had developed a more robust gait – better able to deal with, say, being knocked with a stick (aw!) – than the ones that had learned to walk using upright legs from the beginning.

He reported the results of this experiment – which he said is the first of its kind – in the January 10 online edition of the Proceedings of the National Academy of Sciences. Read more about Bongard’s recent experiment here.

Bongard – – who in 2007 was named a 2007 Young Innovator Under 35 by Technology Review magazine and whose work is supported by the National Science Foundation – conducted this recent experiment as part of a wider venture called evolutionary robotics. He said:

We have an engineering goal to produce robots as quickly and consistently as possible.

Bongard added that humans don’t know how to program robots very well yet, because robots are complex systems. He said that, in some ways, they are too much like people for people to easily understand them.

They have lots of moving parts. And their brains, like our brains, have lots of distributed materials: there’s neurons and there’s sensors and motors and they’re all turning on and off in parallel, and the emergent behavior from the complex system which is a robot, is some useful task like clearing up a construction site or laying pavement for a new road.

Or at least that’s the goal.

But, so far, engineers have been largely unsuccessful at creating robots that can continually perform simple, yet adaptable, behaviors in unstructured or outdoor environments.

Watch this interesting video of Bongard from two months ago, where he talks about the state of robotics now. He mentions two big limitations to having the sorts of robots envisioned b some – say, those in the 2004 film iRobot, which is based on Isaac Asimov’s short story by the same name. In that now-classic film, robots are everyday objects – housemaids, nannies for our children – until they turn bad and Will Smith has to subdue them. In the video below, Bongard mentions two reasons why we’re a long way from the vision of iRobot. First, robots need a power supply. They either have to be plugged into a wall socket, or they have to run on batteries, which run down quickly, as we all know. Second, the real world is a changeable place, and – while robots can be programmed to perform simple tasks – it’s been hard to program a robot to cope with an environment of change.

Still, Bongard and other roboticists are hard at work in their labs. And Bongard’s new research – showing that, like babies, robots have to undergo an evolution in order to learn to walk well – is another step in the direction of a robotic future.

Sure, robots can juggle now. But not all that well.

Watch Jeopardy pit humans vs. artificial intelligence

Deborah Byrd