MITRE Investigates Advances in Motor Control for Next-Generation ProstheticsApril 2011
Topics: Health Innovation
The National Center for Health Statistics estimates that some 133,000 people in the U.S. have undergone an above-the-knee amputation. Traditional, passive prosthetic devices provide mobility, but they often result in a halting gait. This impaired movement can lead to fatigue, pain, and loss of agility.
MITRE researchers are investigating how advances in neuroscience and electronics can be exploited to develop new models for the control of prosthetic devices. These biologically inspired models could lead to the development of next-generation, active prostheses capable of dramatically improving amputee mobility.
"If you're missing a knee, it's difficult to find a replacement that allows for anything approaching natural movement," explains Adam McLeod, a MITRE senior artificial intelligence engineer who works on the project. "With the older prosthetics, you can't go hiking, for example. But the miniaturization of electronics and the increasing availability of cheap sensors are leading to amazing advances in prosthetics research."
Now, instead of being merely passive devices, prosthetics are becoming active. But the capabilities of the latest generation of prosthetics are still a far cry from the astounding range of motion of the biological knee and ankle.
"Newer, powered prosthetics are essentially specialized tools that enable you to perform a specific task, such as walking up stairs," he explains. "But they don't adapt to your needs. They're not a true limb replacement."
A Prosthetic that Follows its User's Lead
MITRE's research proposes that rather than making the patient adapt to the prosthetic, the prosthetic should adapt to the patient.
"Our idea is to reverse-engineer the brain to develop a biologically inspired control strategy for the prosthetic," McLeod says. "This would allow for more natural movement in a cluttered, real-world environment.
"To successfully traverse through the world, you need a large repertoire of movement," he adds. For a prosthetic to help a person do this well, the MITRE team hypothesizes, its functional architecture must mimic the human brain.
"We are interested in mimicking natural movement in challenging environments," says Michael Fine, a MITRE lead artificial intelligence engineer who also works on the project. "The neuroscience behind this is only partially known, but we have a basic understanding of what the underlying architecture of such an advanced model should look like."
Gathering Data from Injured Veterans
To help in their quest to build a better model for prosthetic motor control, Fine and McLeod are working with severely wounded U.S. veterans. "We interviewed veterans who'd had above-the-knee amputations, and who are currently receiving some of the best care available for their injuries," McLeod says. "They share some common frustrations. For instance, active prostheses often pause before a transition, such as switching from walking to stair climbing."
This pause—an "outlier" in terms of the data that fuels the prosthetic's motor control system—means that existing prosthetics present challenges to patients. "Pausing in the subway or in a crowded intersection just isn't an option," says Fine.
"The outliers make all the difference, even in human-made, highly structured environments," McLeod explains.
Reverse-Engineering Human Movements
A baby's ability to step shortly after birth and rare cases of injury where walking can be produced after the spine is transected support the idea that patterns of movement are stored at the base of the spine, Fine explains. This understanding has influenced the emerging design of autonomous robots and intelligent prosthetics. By using a state-of-the art motion capture system to analyze data on how humans walk (see sidebar, "Motion Capture Suit Aids Locomotion Data Collection"), the team is attempting to reverse-engineer those patterns. Then they plan to use those patterns to accurately reconstruct complex movements using a prosthetic.
Biologically inspired control of prosthetics is expected to have important implications for rehabilitation of those with catastrophic injuries. For MITRE's military sponsors, for example, advances in prosthetics could mean a new lease on life for gravely injured U.S. veterans.
But the research also has implications beyond the veteran population, McLeod says.
"The type of control model we are developing for active prosthetics could also be applied to autonomous robots, allowing them to ambulate through a wider variety of environments. When robots can physically reach more locations, they have the opportunity to see and do a wider variety of things, resulting in improved situational awareness and reduction in risk for warfighters," McLeod says.
by Maria S. Lee