Meta's new robotic touch sensors can detect forces smaller than a paperclip

Meta's AI Research team, FAIR, has introduced new technologies that are set to advance touch perception and dexterity in robotics further. They've introduced three standout innovations: Meta Sparsh, Meta Digit 360, and Meta Digit Plexus.

Meta Sparsh is making waves as the first all-purpose system for touch-based sensing using vision—it works across different sensors and tasks. This tech processes touch signals using AI models, skipping over the need for handcrafted models that usually require a lot of customization for each task.

Then there's the Meta Digit 360, which Meta teamed up with GelSight Inc. to develop. This high-tech artificial fingertip boasts over 18 sensing features, can feel forces as tiny as one millinewton, and has more than eight million tiny sensing units (or "taxels") that let it detect touch from any direction.

Adding to the lineup, Meta Digit Plexus is designed as a common platform to bring various touch sensors together on robotic hands. It's all set to be used in the upcoming Allegro Hand from Wonik Robotics, which is expected to come out next year.

Meta has also launched something called PARTNR, a benchmark framework to test how well robots can plan and collaborate with humans. This system includes 100,000 natural language tasks spread across 60 virtual homes and over 5,800 unique objects, all geared toward making robots more socially savvy.

These advances are likely to make a splash across sectors like healthcare, manufacturing, and prosthetics. Digit 360 from GelSight Inc. is set to hit the market next year with an early-access option through a call for proposals.