One of the challenges in robotics is to design a robotic hand that can manipulate objects with the same dexterity and sensitivity as a human hand. A key aspect of this is to enable the robotic hand to identify objects by touch, without relying on vision or multiple regrasps. In this article, we will explore how MIT researchers have developed a robotic hand that can accurately identify objects with just one grasp, using high-resolution touch sensing along the entire length of each finger.

What are the limitations of existing robotic hands?

Most robotic hands use either of two approaches to touch sensing: packing powerful sensors into the fingertips, or spreading lower-resolution sensors along the entire finger. Both approaches have their drawbacks.

The first approach requires an object to be in full contact with the fingertips to be identified, which can take multiple grasps. Moreover, the fingertips are prone to wear and tear, and the sensors can be damaged by sharp or abrasive objects.

The second approach captures less detail about the object’s shape, and can miss important features that are only accessible by the fingertips. Furthermore, the sensors can interfere with the finger’s flexibility and strength, limiting its ability to grasp different objects.

How does the MIT robotic hand overcome these limitations?

The MIT team, led by Professor Edward Adelson, has developed a robotic hand that combines the advantages of both approaches, while avoiding their drawbacks. The robotic hand consists of three fingers, each with a rigid skeleton encased in a soft outer layer that has multiple high-resolution sensors incorporated under its transparent skin.

The sensors use a camera and LEDs to gather visual information about the object’s shape, by capturing the contours that appear on the soft skin when it deforms around the object. The sensors provide continuous sensing along the finger’s entire length, enabling each finger to capture rich data on many parts of an object simultaneously.

Using this design, the researchers built a three-fingered robotic hand that could identify objects after only one grasp, with about 85 percent accuracy. The rigid skeleton makes the fingers strong enough to pick up a heavy item, such as a drill, while the soft skin enables them to securely grasp a pliable item, like an empty plastic water bottle, without crushing it.

What are the applications and benefits of the MIT robotic hand?

The MIT robotic hand has several applications and benefits for various domains and tasks. Some of these are:

– Industrial automation: The robotic hand can handle a wide range of objects with different shapes, sizes, weights, and materials, without requiring prior knowledge or vision. This can improve the efficiency and safety of manufacturing processes, such as assembly, sorting, or packaging.

– Medical robotics: The robotic hand can perform delicate operations that require high precision and sensitivity, such as surgery or rehabilitation. It can also avoid damaging human tissues or organs by applying appropriate force and feedback.

– Service robotics: The robotic hand can interact with humans and their environments in a natural and friendly way. It can assist humans with daily tasks, such as cooking, cleaning, or personal care. It can also provide social and emotional support for people who need it.

Conclusion

The MIT robotic hand is one of the recent developments in robotics that demonstrates how high-resolution touch sensing can enable a robotic hand to identify objects with just one grasp. By combining soft and rigid elements along the entire length of each finger, the robotic hand can mimic human movements, gestures, speech, and facial expressions. communicate with humans using natural language, voice recognition, and speech synthesis. They can also learn from their interactions and adapt to different situations.

—

Source: 

(1) Robotic hand can identify objects with just one grasp. https://news.mit.edu/2023/robotic-hand-can-identify-objects-just-one-grasp-0403.

(2) Robotic hand can identify objects with just one grasp. https://computing.mit.edu/news/robotic-hand-can-identify-objects-with-just-one-grasp/.