Robots on the Move: Boston Dynamics Hits the Brains Over Brawn Switch

Photo by Web Summit | License
Boston Dynamics has been serving looks with its impressive lineup of robots, from agile parkour performers to groovy dancers. While it’s tempting to think that these robots are just for show, founder Marc Raibert is leading a serious charge into the realm of machine learning, hoping to create robots that can do more than just perform party tricks.
Recent innovations mean their robots are starting to learn on their own, which is fancy talk for getting smarter without human input. “The hope is to produce loads of behavior without handcrafting every little thing that robots do,” Raibert revealed. Enjoy running from a robot uprising? Well, get ready!
Boston Dynamics isn’t the only game in town when it comes to cool robots. Other startups are rolling out their own impressive humanoids, all trying to flex their AI muscles. Just this week, Figure unveiled its humanoid grocery unloading wonder called Helix. Meanwhile, x1’s sleek NEO Gamma is busy playing house, and Apptronik is upping production of its Apollo robot. Despite the excitement, it’s hard to ignore the murky pricing and uncertain market for these glorified helpers. Who wants to pay top-dollar for something that might just end up as a high-tech paperweight?
Of course, the real litmus test will be how much these robots can achieve without a human babysitter. Most of their fancy footwork remains reliant on direct programming, but Raibert assures us that a leap towards independence is on the horizon. Last fall, innovations in creating new models for control started gaining traction, which could accelerate robot evolution.
Boston Dynamics offers Spot, a four-legged robot tackling rugged terrain much like a dog runs after a squirrel. Spot can now sprint three times faster thanks to what they call reinforcement learning, a technique that’s rooted in letting AI learn through trial and error, just like us mortals attempt to adult.
The Robotics and AI Institute, launched by Raibert in 2022, is working on upping the intelligence of these machines, with the hope that one day they’ll learn to conquer the dishwashing nightmare. There’s caution ahead, though: while we hope for domestic robot helpers, let’s not forget they need to learn without breaking everything in sight.
So, what do you think, fellow tech enthusiasts? Are we ready for a future where robot pals pick up our socks?
AUTHOR: cjp
SOURCE: Wired