Creating Biomedical Technologies to Improve Health


Science Highlight: March 31, 2010

Powered Robotic Legs - Leaping Toward the Future

According to the National Limb Loss Information Center (NLLIC), 1.7 million people in the United States are living with an amputation, and approximately 300,000 of these are above-the-knee leg amputees. Current lower limb prostheses require amputees to expend significant effort when performing multi-joint-coordinated movements. Normally, human knee and ankle joints generate power during walking and other locomotive functions such as walking up stairs or slopes. Unfortunately, even sophisticated, state-of-the-art leg prostheses do not generate power during movement; instead, these passive devices rely upon ground force effects and mechanical components such as hydraulic valves or sliding joints for proper function. To control the prosthesis, users must make extra movements with their hips and residual limb. During walking, leg amputees can expend up to 60% more metabolic energy compared with a healthy person.

Lower leg amputee walking with the Vanderbilt powered prosthesis. The powered prosthesis reproduces many essential characteristics of healthy biomechanical walking, such as knee flexion during the various phases of walking, resistance of ankle flexion during heel strike, and powered ankle flexion during push-off, which other prostheses do not provide.

Lower leg amputee walking with the Vanderbilt powered prosthesis. The powered prosthesis reproduces many essential characteristics of healthy biomechanical walking, such as knee flexion during the various phases of walking, resistance of ankle flexion during heel strike, and powered ankle flexion during push-off, which other prostheses do not provide.

The limitations of passive devices become even more poignant when amputees face stairs and slopes. Amputees must walk up slopes or stairs one leg at a time, with the prosthetic leg lagging behind. Walking down stairs is even more limiting and potentially dangerous. Individuals with two healthy legs dissipate significant power as they walk down a flight of stairs by stepping toe first, which enables the ankle joint to absorb energy; this prevents excessive momentum when descending the stairs. In contrast, amputees must walk down stairs heel first and often cannot control their acceleration, which can lead to falls. In fact, amputees fall as often as elderly persons living in institutions.

Integrating Man and Machine

In hopes of overcoming these obstacles, Michael Goldfarb, Vanderbilt University professor of mechanical engineering, and co-researchers Huseyin Atakan Varol and Frank Sup have developed a robotic lower leg prosthesis with powered knee and ankle joints. “Only in recent years has robotics technology been adequate to develop a powered leg that could reproduce the biomechanics of a human leg,” says Goldfarb.

“We are entering what some people call the Age of Bionics…a phase where man and machine are becoming integrated,” adds Varol. “I think our work is a pioneering example of this.” Thanks to advances in battery technology, the prosthesis has a completely self-contained battery that can provide power equivalent to muscle and enough torque to replace knee and ankle joint function. “This power solves a number of problems that would increase the quality of life of amputees in general,” says Varol. The battery is rechargeable and can easily last a full day between charges. This equates to approximately 13,000 to 14,000 steps. This is a valuable feature considering the average person walks 7,000 to 12,000 steps daily. The typical amputee walks significantly fewer steps. The Vanderbilt leg has brains and brawn. Dr. Goldfarb and his team have integrated microelectronics technology into the prosthesis, which allows intelligent communication with the user. Unlike most prosthetic devices that cannot adapt or multitask well, the Vanderbilt prosthesis can actively adapt to the user’s environment and movements, as would occur in everyday life. “Our focus is to restore normal gait in everyday living,” says Sup.

Sensors allow the prosthesis to infer what action the user wants to perform, such as moving from a seated position to standing. Sophisticated software, called intent recognizers, behave much like a computer and analyze the user’s movement patterns, like weight shift or changes in joint angles. The patterns are then interpreted into power-driven action that helps the user go from sitting to standing with less effort. The analysis occurs within a few hundred milliseconds and allows smooth integration between the action of the user and the leg. Though the intent recognizer cannot be 100% accurate, secondary checkpoints prevent the user from perceiving any error. “Cognition is the key,” says Varol. “It’s important for the prosthesis to understand what the user wants to do.”

Let Them Walk

The Vanderbilt prosthesis has been tested on only one amputee, but the results have been more than promising. The individual has logged more than 300 hours with the robotic leg and was able to emulate a healthy gait, even when carrying a bag or backpack. His self-selected walking speed was 25% faster than with his previous state-of-the-art leg, and he expended 30–40% less energy. The prosthesis weighs 9.5 pounds, about 2 pounds heavier than his current prostheses; however, the user did not perceive the difference because the prosthesis generates significant energy. More amputees will test the leg in the coming months, and the prosthesis will be optimized for walking down slopes and stairs.

Dr. Goldfarb and his group anticipate their prosthesis will be ready for commercial development by August and ready for market sale in the next 3 to 5 years. Its pricing is expected to be competitive with current state-of-the-art prosthetic legs.

Balancing Act

Dr. Goldfarb and his colleagues also hope to endow the leg with reflexes. For example, if an individual stumbles while walking, the leg will respond to prevent him or her from falling. “That requires the ability to act autonomously, which current prosthetic legs do not have, but our leg does have,” says Goldfarb. Passive prostheses do not assist the user while standing on uneven ground, and so amputees often will effectively stand on only one leg. “Our leg has enough sensors to know what kind of ground it is standing on and provides two-legged balance,” adds Goldfarb. Improving balance and stability for amputees has the most potential for health benefit because it will help prevent falls.

The Bionic Leg

Though the Vanderbilt leg is stronger and smarter than current prostheses, it does have limitations. Its processors only allow it to respond to patterned activity, such as walking or climbing stairs. The leg cannot efficiently recognize non-patterned activities such as dancing or playing basketball. This would require a more complex source of input. Currently, prosthetic legs have no electrical connections to the user; they are attached with a socket and held on by suction. Researchers have begun to look at integrating the nervous system with prosthetic legs. This involves implanting electrodes in the nerves of the amputated leg that can communicate with the joints of the prosthesis. The prosthetic leg could then respond directly to neural commands without conscious thought on the part of the user. Some work is being done in animal models, but the technologies required for neural integration of robotic prostheses do not yet exist. Nonetheless, Dr. Goldfarb and his team are already thinking to the future.

This work is supported in part by the National Institute of Biomedical Imaging and Bioengineering.


Varol HA, Sup FC, Goldfarb M. Multiclass Real-Time Intent Recognition of a Powered Lower Limb Prosthesis. IEEE Transactions on Biomedical Engineering. 2010;57(3):542-51.

Sup F, Varol HA, Mitchell J, Withrow TJ, Goldfarb M. Preliminary evaluations of a self-contained anthropomorphic transfemoral prosthesis. IEEE/ASME Transactions on Mechatronics. 2009;14(6):667–76.

Program Area: 
Health Terms: