Switching from one smartphone to a different is generally a clean process. You log into your accounts and your apps, preferences, and contacts ought to sync to the brand new {hardware}. However on this planet of robotics, swapping an previous robotic arm for a more moderen mannequin has meant setting every thing up from scratch.
To repair that, a workforce of researchers on the Swiss École Polytechnique Fédérale de Lausanne (EPFL) has developed what they name Kinematic Intelligence, a framework that makes switching robots work extra like switching smartphones. They describe their system in a current Science Robotics paper.
Demonstrating expertise
For years, roboticists have been engaged on getting robots to study from demonstration—instructing them new expertise by displaying them what to do, somewhat than writing strains of code. The concept is to remotely management or bodily information the robotic’s arm to show it a process like wiping a desk, stacking containers, or welding a automobile part. The issue is that the majority of those taught expertise find yourself tied to the precise robotic the coaching was completed with.
However robotics is advancing shortly. “The robots have completely different designs, and these days there are new designs being proposed—that brings its personal set of challenges,” mentioned Sthithpragya Gupta, a roboticist at EPFL and lead creator of the examine. If a brand new robotic has barely longer hyperlinks, a unique joint orientation, or a extra advanced configuration, that realized habits immediately breaks and the brand new robotic will doubtless flail, freeze, or crash if making an attempt it.
“With new designs come completely different capabilities and constraints,” mentioned Durgesh Haribhau Salunkhe, an EPFL roboticist and co-author of the examine. “The issue is to adapt to those constraints and capabilities—to faithfully replicate the actions demonstrated by a human.” At this time, making the leap from one robotic physique to a different normally means ranging from scratch and retraining the entire system.
