Regenerative Power

The Cleveland State University: Reimagining prostheses for improved walking

Electric mobility is also an exciting topic in medical technology. Find out how recuperation and advanced control technology help people concerned get ahead.

Lower-limb amputees typically use passive (non-powered) prostheses, which provide support but are unable to produce assistive forces for key activities, including walk- ing and ascending slopes. Because of this, amputees who wear passive prostheses have a significantly larger metabolic energy demand in comparison to able-bodied people. Power prostheses are powered by electric motors. They address the limitations of passive devices by providing a sense of freedom and the ability to walk longer distances with reduced effort. But these devices are limited by their high power consumption. A research team from Cleveland State University and the Louis Stokes Cleveland VA Medical Center received funding from the U.S. National Science Foundation to search for a solution that will greatly expand the range of use and naturalness of motion possible with these devices. The team developed a prosthesis prototype that features energy regeneration technology with supercapacitors as energy storage elements. The team developed a powered knee prosthesis and an advanced energy-optimal control system. It is their hope that the prosthesis will not only ope-rate for a longer period of time but will facilitate a more active lifestyle by enabling faster walking speeds and, eventually, the ability to climb stairs.

Energy Regeneration May Be Key

According to Professor Hanz Richter, the biggest obstacles with these devices are their high electrical demand and the complexity of the actuators that are required to control movement. Today, most electrically powered prostheses only support walking at an average pace and may have to be recharged several times a day. The research team developed a control method that allows for seamless transitions between the different gait phases for walking, including different walking speeds and a range of varying inclining and declining slopes, based on self-modulated impedance control and energy regeneration technologies. Energy regeneration technology has the potential to reduce energy usage in motion systems. This technology is based on the recovery, storage, and reuse of surplus energy associated with motion cycles. Regenerative braking is commonly used in electric vehicles to improve energy efficiency. The same method can be applied to a powered prosthesis. Natural gait involves periods of surplus energy at the knee joint. While passive prostheses act as a brake and dissipate this energy, the regenerative prostheses developed by Prof. Richter can store and reuse surplus power without compromising the naturalness of the motion. Thus, energy regeneration can extend battery life, making a prosthesis more practical for daily use.

Issues with Traditional Powered Prostheses

Most electrically powered prostheses use finite state impedance controllers, mechanical springs, and dampers to replicate the motion of the knee and ankle joints. The prosthesis is divided into a series of gait states, representative of the balance and coordination that is required to complete a walking stride. For each gait state, a separate impedance controller is used. These controllers are triggered by sensors placed on the prosthesis. Control parameters are tuned for each state to match the different walking speeds and walking patterns of individual subjects. Typically, these prostheses have a five-state controller that operates at 3 or 4 different walking speeds and the controller parameters require tedious impedance scheduling. For instance, 5 walking gait phases with 3 gains each and 3 speeds imply 45 gains to be tuned. In contrast, the team used a continuous impedance modulation scheme based on axial shank force, allowing for a substantially reduced control parameter space.

Test subject walking with the prosthesis prototype.

Issues with Traditional Powered Prostheses

Most electrically powered prostheses use finite state impedance controllers, mechanical springs, and dampers to replicate the motion of the knee and ankle joints. The prosthesis is divided into a series of gait states, representative of the balance and coordination that is required to complete a walking stride. For each gait state, a separate impedance controller is used. These controllers are triggered by sensors placed on the prosthesis. Control parameters are tuned for each state to match the different walking speeds and walking patterns of individual subjects. Typically, these prostheses have a five-state controller that operates at 3 or 4 different walking speeds and the controller parameters require tedious impedance scheduling. For instance, 5 walking gait phases with 3 gains each and 3 speeds imply 45 gains to be tuned. In contrast, the team used a continuous impedance modulation scheme based on axial shank force, allowing for a substantially reduced control parameter space.

Energy-Regenerative-Powered Transfemoral Prosthesis

The research team developed a prototype to demonstrate their ideas on energy regeneration and self-modulated impedance control. Their prototype consists of a passive ankle and a powered knee joint. The knee joint is actuated by a DC motor with a leadscrew and a crank-slider mechanism. An ultracapacitor (also known as a supercapacitor) is used as an energy storage element instead of a battery, providing an efficient means of storing and reusing energy. The ultracapacitor is lightweight and durable and has high power densities and the ability to rapidly charge and discharge without damage, in contrast to batteries. Another key design element of their prototype is the control method they have devised. The team developed a novel varying-impedance control ap- proach that drives the prosthesis in both the stance and swing phase, while explicitly dealing with energy regeneration. The control method varies the impedance of the knee joint based on the amount of force exerted on the shank and promotes energy regeneration by precisely injecting a designated amount of negative damping into the system. “This approach provides a natural variation in the impedance of the knee and leads to far fewer tuning para-meters compared to some other approaches,” said Professor Richter. “Additionally, the controller allows walking at different speeds without the need for retuning. With a simple adjustment, the same tuning can be used for different subjects.”

Validating the Control Method

To validate their control method, the team installed a variety of sensors on the test prosthesis to gather data on the control method and to evaluate its overall performance. A volunteer amputee was recruited, and trials were conducted. To obtain feedback on the control strategy, the motor position, which is kinematically related to the knee angle, was measured by an encoder to compute velocity. Additionally, two strain gages were installed and then calibrated to produce shank force measurements. The voltage of the ultracapacitors was then measured for use as feedback as part of the semiactive virtual control method developed by the team. To evaluate the energy regeneration capacity of the prosthesis, sensors were installed on both sides of the motor driver to measure input and output currents. The voltage applied to the motor, as well as the ultracapacitors, were recorded. The combined measurements provided information on the overall power usage and the efficiency of the motor driver. A dSPACE system was used for the centralized data acquisition, control, and display for the prosthesis and its control system, at a rate of 1 kHz. Real-time control computation was implemented in a Simulink block diagram where some code was executed by embedded MATLAB® blocks. Digital filtering with a 24 Hz cutoff frequency was applied to all measurements. A tether cable was connected between the user and the dSPACE system. According to Prof. Richter, the dSPACE system and Simulink compatibility greatly helped the team focus on the control algorithms, rather than on the implementation details. In particular, the transition from modeling and simulation to the real-time deployment were streamlined.

Trials on a Treadmill

After completing initial validation tests, the team began a series of human trials. During tests with a 35-year-old male amputee walking at three different speeds on a treadmill – slow (0.6 m/s), preferred (0.75 m/s) and fast (0.9 m/s). Through this trial, the control method was validated and energy regeneration was achieved under the test conditions. A 10-camera passive marker motion capture system recorded 26 markers placed on standard anatomical locations. The force plates of the split belt treadmill measured ground reaction forces for each side.

Prosthetic knee prototype.

Trials on a Treadmill

After completing initial validation tests, the team began a series of human trials. During tests with a 35-year-old male amputee walking at three different speeds on a treadmill – slow (0.6 m/s), preferred (0.75 m/s) and fast (0.9 m/s). Through this trial, the control method was validated and energy regeneration was achieved under the test conditions. A 10-camera passive marker motion capture system recorded 26 markers placed on standard anatomical locations. The force plates of the split belt treadmill measured ground reaction forces for each side.

Regenerated Energy provides Power to the Knee Joint

In summary, the team observed that the tuning process was relatively easy – tuning was completed in a matter of minutes while conducting the test. They also observed energy regeneration taking place, providing power to the knee joint when needed. However, the team realized that further improvements in energy regeneration are possible. Though the project has concluded, additional funding would allow the team to focus on the stability of the controller and improve energy losses that were detected in the first prototype. The data collected from the trials for a human-side evaluation and the results will be featured in an upcoming full-length paper in the Medical Engineering and Physics Journal.

About the author:

Professor Hanz Richter

Professor Hanz Richter

Department of Mechanical Engineering Cleveland State University

S'inscrire à la Newsletter

S’abonner à nos newsletters, gérer ses abonnements ou se désabonner