Skip to content
AquiaRobotics

Story 01 · 12 October 2026

What happened to the robot hand with no fingers?

A soft bag filled with grains can grip an object when air is drawn out. Researchers made it work. Turning that idea into a product brought a different set of problems. We follow the experiment and the attempt to commercialise it.

Field of viewOne paper, its authors’ own account of their company, and the patents and trade-show lists in our record. One case: it shows what commercialising one soft gripper took, not how often research reaches a product.How much can this map see?

Measured

01 · The grip

How do you hold something without fingers?

1 · meets2 · settlesair out3 · holds
A schematic, not a measurement. A rubber membrane full of loose grains meets an object (left); while the grains flow it settles around the shape (middle); with the air drawn out the grains lock and the membrane holds (right). Drawn from the description in Brown and colleagues, 2010, and Amend and colleagues, 2016.Measured

In 2010 a paper described a gripper with no fingers: a rubber bag of loose grains — ground coffee will do — pressed onto an object. While the grains can flow, the bag settles around the shape. Draw the air out and the grains lock together; the bag goes hard and holds. Let the air back in and it lets go. It looks more like a beanbag than a hand. The paper of 2010 was the first to make it work and measure it, and it travelled: cited 1,560 times as OpenAlex counts, 1,352 of those in our collection. Then came the harder question, which papers rarely answer: could you buy one?

02 · The product

Who tried to sell it, and what did they have to change?

The lab's balloon1000Shipped, sharp objects5000Shipped, on average50,000Shipped, smooth objects90,000grips before the membrane fails
How many times a membrane could grip before it failed, as the company measured it on its own test machines: the party balloon of the laboratory prototypes, then the polychloroprene membrane it shipped, which wore faster on sharp objects and slower on smooth ones. Source: Amend and colleagues, 2016, ‘Membrane durability’.Measured
Data table
Itemgrips before the membrane fails
The lab's balloon1000
Shipped, sharp objects5000
Shipped, on average50,000
Shipped, smooth objects90,000

Two Cornell students — John Amend, an author of those papers, and Bill Culley — founded Empire Robotics in 2012 to sell the gripper to factories, and sold it from 2014 as the VERSABALL. Their own account, published in 2016, lists the questions a factory asks a gripper before buying it — from air consumption per grip to shelf life — and says which ones decided sales. The lab’s balloon lasted about 1,000 grips. A line making three grips a minute would wear one out in days, so the wear parts went into a head that snaps off, and the company learned to dip-mould its own membranes: more than 50 rubber formulas to reach about 50,000 grips.

03 · The outcome

Why did the company close?

Research grants$1.5 MPrivate investment$500 kSales$500 kUS dollars over four years, as the authors state them
What four years produced, in the authors’ own figures: grants won, money raised from investors, and revenue from sales — ‘many to major companies’. Alongside, and not drawn: more than 100 customer applications tested, 350 more potential customers interviewed, at most 10 employees, more than 20 trade shows. Source: Amend and colleagues, 2016.Measured
Data table
ItemUS dollars over four years, as the authors state them
Research grants$1.5 M
Private investment$500 k
Sales$500 k

In spring 2016 Empire Robotics closed. The authors’ explanation, in their own words: they could not “identify and solve a large enough market problem with the product specifications we were able to achieve”. Not for want of trying. Interest came overwhelmingly from factories — plastics 25%, consumer goods 20%, car parts 15%, packaging 15%, food 10% — and the gripper was sold, shown at more than 20 trade shows, tested on more than 100 applications. Read the account for what it is: one document, written by people with a financial interest in the company, explaining a technical ceiling. It does not say what those customers bought instead, or what a competitor offered, and neither do we.

04 · What we learned

What does one gripper teach about turning research into a product?

air out3 · holdsthe two heads, to scale9 cm16.5 cm
The jammed grip once more, beside the product’s two heads drawn to scale: 9 cm and 16.5 cm in diameter, the two sizes the company sold. A schematic, not a measurement; the diameters are the authors’ own.Measured

A gripper that works in a laboratory is a membrane and some grains. A gripper that works in a factory is a head that snaps off, a membrane that survives 50,000 grips, two sizes, a pump, a filter and a price. Two head sizes, 9 cm and 16.5 cm, because a gripper holds best an object about 50% of its own diameter. The distance between the two is not the idea; it is everything around it — and that distance is what this account measures. What happened to the hand with no fingers: it was built, sold and withdrawn, and the mechanism is still in the papers and in the patents.

A note on where the name went next. John Amend’s name appears on 68 US robotics patents in our record, the first owned by Cornell University, the later ones by Berkshire Grey, a warehouse-robotics company. That is a name on patents, not this gripper: the company’s own patent list names four products and their patents, and we have not checked whether any of them is a jamming gripper. The company exhibited at MODEX in 2026 — evidence of exhibiting, nothing more.