Tech

Watch researchers bully this robot dog for science

Just don't let our future robot overlords see this one.

Teaching robots how to walk has been a long road, but recent progress from pioneers like Boston Dynamics, the company responsible for bots like Spot and Atlas, have made agile humanoid and canine-like machines a reality.

And while robots are better than ever at mimicking human locomotion (see this clip of robo-parkour) that doesn't mean they're infallible. In fact, for robots, just like a human learning how to move, it's not a matter of if, but when they take a spill, which is why training them how to recover after a fall is just as important.

There's just one problem: recovery often involves lots of unique scenarios that can be difficult to fully encode into a bot's repertoire ahead of time. That's exactly why researchers from Zhejiang University and the University of Edinburgh have tasked themselves with making software that allows robots to teach themselves on the fly.

In a paper recently published in the journal Science Robotics, Multi-Expert Learning Architecture (MELA) which uses multiple neural networks to compute situations that the AI has not necessarily encountered before.

As noted by Wired, MELA consists of eight separate "expert" algorithms that are trained for different tasks like walking or maintaining balance. Those are all managed by a separate overarching "coach" network that decides which skill to pull upon in any given scenario.

Using this hierarchal system, researchers were able to observe their canine-esque robot named Jueying gather itself into various never-before-seen poses after being knocked to the ground.

The goal is to eventually be able to design robots that are more dynamic and adaptable, allowing them to more easily traverse new terrain.

There are non terrestrial obstacles before MELA makes its way into real-life robots, however. The system, while effective, also uses a great deal of computing power to simulate training, making it difficult to apply MELA in more casual, non-research settings.

With added efficiency and refinement, however, it wouldn't be surprising to see MELA show up in other robots outside of Jueying. Let's just hope that by the time those robots are in charge they don't see the behind-the-scenes footage.

Share