Rocket-powered Mechanical Arm Could Revolutionize Prosthetics

A prototype of this radical design has been successfully developed and tested by a team of mechanical engineers at Vanderbilt University as part of a $30 million federal program to develop advanced prosthetic devices.

"Our design does not have superhuman strength or capability, but it is closer in terms of function and power to a human arm than any previous prosthetic device that is self-powered and weighs about the same as a natural arm," says Michael Goldfarb, the professor of mechanical engineering who is leading the effort.

The prototype can lift (curl) about 20 to 25 pounds — three to four times more than current commercial arms — and can do so three to four times faster. "That means it has about 10 times as much power as other arms despite the fact that the design hasn't been optimized yet for strength or power," Goldfarb says.

The mechanical arm also functions more naturally than previous models. Conventional prosthetic arms have only two joints, the elbow and the claw. By comparison, the prototype's wrist twists and bends, and its fingers and thumb open and close independently.

The Vanderbilt arm is the most unconventional of three prosthetic arms under development by a Defense Advanced Research Project Agency (DARPA) program. The other two are being designed by researchers at the Advanced Physics Laboratory at Johns Hopkins University in Baltimore, who head the program. Those arms are powered by batteries and electric motors. The program is also supporting teams of neuroscientists at the University of Utah, California Institute of Technology and the Rehabilitation Institute of Chicago who are developing advanced methods for controlling the arms by connecting them to nerves in the users' bodies or brains.

"Battery power has been adequate for the current generation of prosthetic arms because their functionality is so limited that people don't use them much," Goldfarb says. "The more functional the prosthesis, the more the person will use it and the more energy it will consume."

Increasing the size of the batteries is the only way to provide additional energy for conventionally powered arms and, at some point, the weight of additional batteries becomes prohibitive.

It was the poor power-to-weight ratio of batteries that drove Goldfarb to look for alternatives in 2000 while he was working on a previous exoskeleton project for DARPA. He decided to miniaturize the monopropellant rocket motor system that is used for maneuvering in orbit by the space shuttle. His adaptation impressed the Johns Hopkins researchers, so they offered him $2.7 million in research funding to apply this approach to the development of a prosthetic arm.

Goldfarb's power source is about the size of a pencil and contains a special catalyst that causes hydrogen peroxide to burn. When this compound burns, it produces pure steam. The steam is used to open and close a series of valves. The valves are connected to the spring-loaded joints of the prosthesis by belts made of a special monofilament used in appliance handles and aircraft parts. A small sealed canister of hydrogen peroxide that easily fits in the upper arm can provide enough energy to power the device for 18 hours of normal activity.

The first prototype, which took a year to develop, was powered by "cold gas": compressed nitrogen. It allowed the researchers to test the fundamental design and to address the basic problems of control, leakage and noise. The team was happy to discover that they could solve all of the basic problems by designing the valves with the highest precision possible, with clearances of 50 millionths of an inch.

"There are only a handful of machinists who can make valves with this precision. We found one and asked him to make them with the highest precision possible, which is actually higher than he can measure," says Goldfarb. "Normally in projects like this the surprises are unpleasant, but this was a pleasant one. The valves didn't leak, click or hiss!"

After getting the arm working with cold gas, the engineers tore it down and rebuilt it to operate on "hot gas" — steam that is heated to 450 degrees Fahrenheit by the hydrogen peroxide reaction.

One of their immediate concerns was protecting the wearer and others in close proximity from the heat generated by the device. They covered the hottest part, the catalyst pack, with a millimeter-thick coating of a special insulating plastic that reduced the surface temperature enough so it was safe to touch. The hot steam exhaust was also a problem, which they decided to handle in as natural a fashion as possible: by venting it through a porous cover, where it condenses and turns into water droplets. "The amount of water involved is about the same as a person would normally sweat from their arm in a warm day," Goldfarb says.

To allow for thermal expansion, the engineers replaced the arm's nine valves with a set machined to a slightly lower tolerance, approximately 100 millionths of an inch. But when they began operating the rebuilt arm, they found that it hissed and leaked. At first, they thought that the arm had only a single leak, and spent several weeks trying to track it down. Finally, they realized that the noise and leakage were coming from all the valves. Replacing the high-precision valve set solved the problem. "We were astonished at by the difference between 50-millionths and 100-millionths: It made all the difference in the world," says Goldfarb.

Their biggest problem operating with hot gas turned out to be finding belt material that was strong enough and could withstand the high temperatures involved. They tried silk surgical sutures, but found that silk wasn't strong enough. They tried nylon monofilament, which is stronger than steel, but it couldn't take the heat. Finally, after a long process of trial and error, they found a material that works: the engineering thermoplastic polyether ether ketone(), .

The engineers solved these and a number of other smaller problems and got the second prototype working properly by the end of June.

In the fall, DARPA's "Revolutionizing Prosthetics 2009" program will move to its second stage. Even though his team has met all its research milestones and has produced a working prototype, Goldfarb is not certain that it will be funded for the new stage. "DARPA has set a goal of developing a commercially available arm in two years. Because of our novel power source, the process of proving that our design is safe and getting regulatory approval for its use will probably take longer than that," he says.

If DARPA decides it cannot continue supporting the arm's development for this reason, Goldfarb is confident that he can get alternative funding. "We have made so much progress and gotten such positive feedback from the research community that I'm certain we'll be able to keep going," he says.