first_imgStay on target Ask a human to hang a mug by its handle on a hook, and they won’t hesitate.Ask a robot to carry out the same task, and you’ll be waiting a long while.Unless you’re at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), where researchers developed an automated system of pick-and-place tasks.“Whenever you see a robot video on YouTube, you should watch carefully for what the robot is NOT doing,” MIT professor Russ Tedrake, senior author of a paper about the project, said in a statement. “Robots can pick almost anything up, but if it’s an object they haven’t seen before, they can’t actually put it down in any meaningful way.”The two most common approaches to picking up objects are “pose-based” systems that estimate an object’s position and orientation, and general grasping algorithms.But these methods, according to MIT, are flawed: Pose estimators often don’t work with objects of significantly different shapes, and grasping approaches can’t place objects with much subtlety.In contrast, CSAIL’s kPAM (Keypoint Affordance Manipulation) approach detects a collection of coordinates (keypoints) on an object—establishing a sort of visual roadmap.Keypoints naturally handle variation among a particular type of object, like a mug or shoe; these coordinates provide all necessary information for a robot to determine what to do with an item.Trial and error (via MIT Computer Science and Artificial Intelligence Laboratory)“Understanding just a little bit more about the object—the location of a few key points—is enough to enable a wide range of useful manipulation tasks,” Tedrake said.In the case of a mug, the system requires only three keypoints—the center of the mug’s side, bottom, and handle. For a shoe, kPAM needs six points to pick up more than 20 different pairs of footwear, from slippers to boots.“This particular representation works magically well with today’s state-of-the-art machine learning perception and planning algorithms,” Tedrake added.Moving forward, the team hopes to teach their system to perform tasks with even greater generalizability—like unloading a dishwasher or wiping kitchen counters.More on Geek.com:Meet the Buddhist Robot That Gives Sermons at an Ancient Japanese TempleThis Weird Robot Can Fold Your Laundry at a Snail’s PaceMIT Removes Inherent AI Facial Recognition Bias Evan Rachel Wood Just As Disturbed by Humanoid Sophia As Everyone ElseMIT’s Thread-Like Robot Slides Through Blood Vessels In the Brain last_img