Switching from one smartphone to another is mostly a seamless process. You log into your accounts and your apps, preferences, and contacts should sync with the new hardware. But in the world of robotics, replacing an old robotic arm with a new model means setting everything up from scratch.
To fix this, a team of researchers at the Swiss Ecole Polytechnique Fédérale de Lausanne (EPFL) has developed what they call kinematic intelligence, a framework that makes switching robots work like switching smartphones. They describe their system in a recent Science Robotics paper.
display of skills
For years, roboticists have been working on making robots learn by demonstration – teaching them new skills by showing them what to do rather than by writing lines of code. The idea is to remotely control or physically direct a robot’s arm to teach it tasks like wiping tables, stacking boxes or welding car components. The problem is that most of these taught skills are tied to the specific robot the training was done with.
But robotics is advancing rapidly. “Robots have different designs, and new designs are being proposed nowadays – which bring their own set of challenges,” said Sthirpragya Gupta, a roboticist at EPFL and lead author of the study. If a new robot has slightly longer links, a different joint orientation, or a more complex configuration, that learned behavior immediately breaks down and the new robot will likely stumble, stall, or crash when attempted.
“With new designs come different capabilities and constraints,” said EPFL roboticist and study co-author Durgesh Haribhau Salunkhe. “The problem is to optimize these constraints and capabilities – to faithfully replicate the functions performed by humans.” Today, jumping from one robot body to another usually means starting from scratch and retraining the entire system.
<a href