Google researchers are using imitation learning to teach autonomous robots how to pace, spin, and move in more agile ways.
What they did: Using a data set of motion capture data recorded from various sensors attached to a dog, the researchers taught a quadruped robot named Laikago several different movements that are hard to achieve through traditional hand-coded robotic controls.
How they did it: First, they used the motion data from the real dog to construct simulations of each maneuver, including a dog trot, side-step, and … a dog version of classic ’80s dance move, the running man. (The last one was not, in fact, performed by the real dog itself. The researchers manually animated the simulated dog to dance to see if that would translate to the robot as well.) They then matched together key joints on the simulated dog and the robot to make the simulated robot move in the exact same way as the animal. Using reinforcement learning, it then learned to stabilize the movements and correct for differences in weight distribution and design. Finally, the researchers were able to port the final control algorithm into a physical robot in the lab—though some moves, like the running man, weren’t entirely successful.
Why it matters: Teaching robots the complex and agile movements necessary to navigate the real world has been a long-standing challenge in the field. Imitation learning of this kind instead allows such machines to easily borrow the agility of animals and even humans.
Future work: Jason Peng, the lead author on the paper, says there are still a number of challenges to overcome. The heaviness of the robot limits its ability to learn certain maneuvers, like big jumps or fast running. Additionally, capturing motion sensor data from animals isn’t always possible. It can be incredibly expensive and requires the animal’s cooperation. (A dog is friendly; a cheetah, not so much.) The team plans to try using animal videos instead, which would make their technique far more accessible and scalable.
To have more stories like this delivered directly to your inbox, sign up for our Webby-nominated AI newsletter The Algorithm. It's free.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Unpacking the hype around OpenAI’s rumored new Q* model
If OpenAI's new model can solve grade-school math, it could pave the way for more powerful systems.
Minds of machines: The great AI consciousness conundrum
Philosophers, cognitive scientists, and engineers are grappling with what it would take for AI to become conscious.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.