Sphinx helps robots wrest control of precision tasks in tight spaces. Yale University
Robots can lift heavy loads and work nonstop, tirelessly moving boxes or assembling parts on an assembly line.
But ask them to perform something simple for a human, like turning a door handle, twisting open a jar, or unscrewing a light bulb, and they will often fumble awkwardly.
Precise maneuvers, especially when working in tight or unpredictable spaces. Until now, this limitation has been a major barrier to designing machines that can seamlessly transition from factory floors to everyday environments.
Scientists at Yale University have found a way to fix this long-standing problem, giving robots the ability to handle far more complicated movements with greater ease and efficiency.
Their innovation could help bridge the gap between industrial robots and the kind of adaptable, nimble machines needed in homes, hospitals, and disaster zones.
Typically, robots use a gripper paired with a wrist that has three degrees of freedom: roll (rotating front to back), pitch (side to side), and yaw (vertical movement).
But these wrists are mechanically complex and often positioned far from the object being held, forcing the robot to move its entire arm to make adjustments. The result is awkward, inefficient motions that also take up more space.
Yale researchers in Prof. Aaron Dollar’s lab have developed a robotic hand, nicknamed the Sphinx, that tackles this problem.
Its spherical mechanism can both grasp and rotate objects along all three axes, combining much of the functionality of a traditional wrist and gripper into a single, streamlined design.
“It’s not very complex,” Vatsal Patel, lead author of the paper and a Ph.D. candidate in Dollar’s lab, said.
“It doesn’t have any sensors or anything on it. It works without any cameras or sensors and things like that. But because of the spherical mechanism, it’s always going to roll, pitch, and yaw objects.”
“It’s a lot more efficient, and you don’t need a ton of space. The wrist is able to do these rotations much closer to the object without the burden of moving the whole arm. It works much faster and much more efficiently.”
The new design significantly improves a robot’s ability to operate in tight or cluttered spaces.
For example, robots can now perform a task as delicate as screwing in a lightbulb inside a cramped closet with great ease.
Beyond such specific scenarios, this advancement moves the field closer to a long-standing ambition of creating robots that can seamlessly navigate and perform tasks in complex, unstructured environments, whether that’s assisting in homes, maneuvering through disaster zones, or adapting to unpredictable real-world settings.
“In these environments, robots don’t know exactly where the objects are,” Patel said.
“They’re trying to adapt to the environment, adapt to the objects. That’s where robotics is generally moving, and we’re trying to solve the same problems.”
The findings of the study have been published in the journal Nature Machine Intelligence.
MasterCard