Researchers at the École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland have developed a robotic arm capable of processing and catching moving objects in just a fraction of a second. The team, that works at the institute’s Learning Algorithms and Systems Laboratory (LASA), was able to teach the robot to understand the path of the object and intercept it at blistering speed. As well as being extremely cool to watch, it’s possible that the technology might find safety-oriented applications in the future.
In order to teach the robotic arm to catch the moving objects, the team used a method known as "programming by instruction." Also utilized for teaching the Baxter industrial robot how to perform certain tasks, the technique involves showing the robot various possible trajectories that the object in question might take, and then manually and repeatedly moving the 1.5-m (4.9-ft) arm to the point at which it would catch the projected target. This method, which was inspired by the way that humans themselves learn, effectively allows the robot to build a "how to catch" model that it can then apply in practice.
The robot uses several cameras located around it to build a real-time 3D model, and in trials was able to catch objects including a half-full bottle of water, a hammer and a tennis racket, within less than five hundredths of a second. The asymmetric nature of the objects, as well as the necessity to catch some of them at certain points (for example, the handles on the hammer and the racket), presented a significant challenge to the robot, making its catching abilities all the more impressive.
Team member Aude Billard commented on the potential uses of the technology, stating that it "can be used for many different other applications, typically catching people that are in danger of falling [or] catching an object that may fall onto people." As such, it’s easy to imagine the technology finding a place in dangerous environments such as construction sites.
To see the EPFL’s catching robot in action, check out the video below.
See the stories that matter in your inbox every morning