Why do human children learn to walk so much easier than robots do? Because babies learn to crawl before they walk. University of Vermont robotics expert Josh Bongard demonstrates how robots develop faster when their bodies change with their brains.
Walking is harder than it looks. Do a YouTube search for walking robots and you’ll see all sorts of slow-moving, oddly-gaited machines attempting to mimic human movement and failing. There’s a complexity to the way our muscles and joints work together to propel us forward that is hard to translate into mechanical minds. Funny that it’s so simple even a child can do it.
The key to our mastery of moving on two legs might stem from the way we develop. Babies are born without the capability of standing on their own two feet, so they first learn to crawl. This teaches them the basics of locomotion, and by the time they’re ready to make their first two-legged trips they’ve already got that down.
It’s a concept that Josh Bongard now applies to robots. Most robotics engineers aim at creating a robot capable of walking and then teaching it to do so. Bongard is creating robots without the capacity for walking and teaching them to crawl first, demonstrating how morphological change can help design better robots. The results of his study were published in the January 10 edition of the journal Proceedings of the National Academy of Sciences.
The test this concept, Bongard created a sophisticated computer simulation in which generations of simple robots ran a genetic algorithm that saw them testing out various methods of locomotion in order to reach a light placed at the other end of a virtual plain. The key is for the robots to reach the light without tipping over.
He found that robots that started off as snake-like creatures unable to stand upright and evolved in stages into robots capable of moving on all four legs learned to move much faster than those with full capabilities from the start.
“The snake and reptilian robots are, in essence, training wheels,” says Bongard, “they allow evolution to find motion patterns quicker, because those kinds of robots can’t fall over. So evolution only has to solve the movement problem, but not the balance problem, initially. Then gradually over time it’s able to tackle the balance problem after already solving the movement problem.”
Once the computer simulation was finished, Bongard put his findings to the test in the real world.
“We built a relatively simple robot, out of a couple of Lego Mindstorm kits, to demonstrate that you actually could do it,” he says. This physical robot is four-legged, like in the simulation, but the Lego creature wears a brace on its front and back legs. “The brace gradually tilts the robot,” as the controller searches for successful movement patterns, Bongard says, “so that the legs go from horizontal to vertical, from reptile to quadruped.
“While the brace is bending the legs, the controller is causing the robot to move around, so it’s able to move its legs, and bend its spine,” he says, “it’s squirming around like a reptile flat on the ground and then it gradually stands up until, at the end of this movement pattern, it’s walking like a coyote.”
Bongard’s study could change the way we develop robots. Traditionally robotics engineers have focused on fixed body types, changing behaviour through programming. Bongard says that methods overlooks a key fact about natural development and evolution.
“One thing that has been left out all this time is the obvious fact that in nature it’s not that the animal’s body stays fixed and its brain gets better over time,” he says, “in natural evolution animals bodies and brains are evolving together all the time.” A human infant, even if she knew how, couldn’t walk: her bones and joints aren’t up to the task until she starts to experience stress on the foot and ankle.
Perhaps the Japan-developed M3 Neony humanoid baby robot (seen above) is a step in the right direction?