Yale scientists use 3D printing and urethane casting to create human-like robotic Openhand

While the problem isn’t so prevalent with small 3D printed robots, anyone who has ever worked with a large robotic grabbing setup will have encountered the same frustrating problem: how do you get the end effector (the robot’s ‘hand’) to function like its human counterpart?

While several sensible solutions have already been found, these are typically very expensive, almost impossible to customize and can’t be easily transferred to other setups. If anything, it emphasizes that the human hand is a truly marvellous creation that took millions and millions of years to perfect.

But Yale wouldn’t be Yale if it scientists were so easily put off. Under the leadership of professor of Mechanical Engineering Aaron Dollar, the Openhand Project has come up with a great 3D printed solution. Their hand design is not only remarkably easy and cheap to recreate, it also results in a very functional hand that can firmly grasp a variety of small and large objects – even coins lying flat down on the table. And what’s more, its open source.

As the Yale researchers reveal, their project was started to fuel a larger user community that would enable robotic design to evolve organically. ‘This project intends to establish a series of open-source hand designs, and through the contributions of the open-source user community, result in a large number of useful design modifications and variations available to researchers,’ they write. First on the agenda: robotic hands, which need to be multipurpose and suitable for repeated functional use. ‘Hands developed through this project are therefore designed to be minimalistic and rugged, especially appropriate for iterative design and operation in unstructured environments.’

And while minimalistic, their solution is definitely suitable for repeated use. They have come up with a design that revolvers around tendon-driven underactuated fingers. ‘Underactuated hands have been shown to improve the generality of simple grippers by adaptively conforming to the surface of objects without the explicit need for sensors or complicated feedback systems. This design paradigm separates the actuation and finger elements, enabling a greater degree of customization.’

But while those designs can be very expensive, the Yale team have come up with a manufacturing technique revolving around resin casting, 3D printing and a bit of hands-on work. Essentially, they 3D print a series of plastic fingers which double as resin molds. These hollow structures are filled with resin, which forms the tendons and fingerpads after curing and a bit of cutting. The entire creation process can be seen in the video below.

The result are a series of multi-material fingers that are not only robust and flexible, but also require a minimal number of fasteners and other components to function. These tendon-driven fingers are capable of naturally curling around objects, meanwhile evenly distributing its force arcross the object like a human hand. The fingerpads, furthermore, also enable the hand to pick up flat and small objects with ease, as is shown in the video below

And, as a crowning cherry, both the open source software and CAD files for these tendon-driven fingers have been designed to allow for variable configuration. ‘[This allows] users to quickly change functional parameters (ie. link lengths, transmission ratios) and manufacturing parameters (ie. shell thicknesses, hole dimensions) and have those changes propagate across all relevant parts,’ the Yale team writes. This means that this whole approach to tendon-driven robotic end effectors can be relatively easily adapted to your own build.

Some alternative applications of these 3D printed fingers

In fact, to illustrate the flexibility of these designs, the Yale team has already developed four different end effector designs based on this principle. They have also designed a basic wrist coupling system that perfectly aligns to these tendon-grabbing fingers. If you’d like to try this 3D printing/mold casting approach to grabbing robotics, you can download all the free files here. That link also includes some python libraries that can be used to modify the CAD files to suit your own plans. For more on their premade hand designs and wrist coupling system, go here.