Our hands are pretty incredible appendages. They can pick up a ceramic mug in one hand and a delicate piece of cake in the other – both with equal ease and almost no thought to the matter.
Robots on the other hand, haven’t managed to master the second one. Soft, delicate objects require an exact understanding of what the object is, embedded sensors, complex feedback loops, or advanced machine learning algorithms if these machines are to be able to successfully pick them up.
“Securely grasping an object typically requires some knowledge of its size, shape, and mechanical properties. In the natural world, this is done, seemingly without effort, [for example] by elephants whose trunks can pick up a peanut or uproot a tree,” researchers from Harvard University write in their new paper, published in PNAS.
“In the engineered world of robotic grasping, much work has focused on understanding how to design the mechanics and dynamics of gripper hardware, as well as how to control such devices to interact with objects in the desired way.”
To avoid having to understand exactly what the object is, and create perfectly co-ordinated fine feedback control before the item is grasped, new research has created a robot which uses soft filaments that tangle around the object instead.
Think of this like a mix between the tentacles of a jellyfish, and a hairbrush getting stuck in curly hair.
“With this research, we wanted to reimagine how we interact with objects,” said Kaitlyn Becker, lead author and microrobotics researcher at Harvard.
“By taking advantage of the natural compliance of soft robotics and enhancing it with a compliant structure, we designed a gripper that is greater than the sum of its parts, and a grasping strategy that can adapt to a range of complex objects with minimal planning and perception.”
The new type of soft, robotic gripper uses thin tentacles to entangle and ensnare objects, similar to how jellyfish collect stunned prey. Alone, individual tentacles, or filaments, are weak. But together, the collection of filaments can grasp and securely hold heavy and oddly shaped objects. The gripper doesn’t require sensing, planning, or feedback control.
This is done through inflation. The filaments or tentacles are hollow, rubber tubes that can pressurise and curl up. The curls knot and entangle with each other and the object, with each entanglement increasing the strength of the hold. While the collective hold is strong, each contact is individually weak and won’t damage even the most fragile object. To release the object, the filaments are simply depressurised.
This is a different design but a similar solution to an octopus tentacle-like glove with suckers designed by a team at Virginia Tech, although that was built to be used to help humans, not robots.
Although it’s still in its early stages, this pink tangled mess might one day be used in real-world applications to grasp soft fruits and vegetables for agricultural production and distribution, delicate tissue in medical settings, even irregularly shaped objects in warehouses, such as glassware.
“This new approach to robotic grasping complements existing solutions by replacing simple, traditional grippers that require complex control strategies, with extremely compliant, and morphologically complex filaments that can operate with very simple control,” said Harvard roboticist Professor Robert Wood. “This approach expands the range of what’s possible to pick up with robotic grippers.”