Mechanomyography in powered knee prosthesis control

Ashwin Needham

Industry focus has shifted in recent years to developing effective neural interfaces for people who wear lower limb prostheses. Traditionally leg joint systems have been designed to work independently of the user’s intent, interpreting their desires on their own with no direct user control. Whilst enabling technologies are nascent or theoretical at the current time, interim efforts have led to the use of biological myographic methods. Myoelectric or EMG systems are mature and have been explored for multiple applications for many years now. Whilst commonly used in upper limb applications, translation to lower limb applications has been resisted by limiting factors, largely practical issues, so we are exploring the use of mechano-myography, MMG, which instead listens in on physical activity of the user’s muscles in interpreting their gait intent, stride by stride and moment by moment. By observing muscle activation of the user’s residual upper leg (quadriceps and hamstrings largely) we are hoping to create a plug in module that can infer actuation of a powered knee prosthesis. We are working with Ӧssur, Iceland in advancing the operation of their Power Knee II, a proprietary device that is a market leader across the world and is also unique in the marketplace as a powered knee device, meaning it can power the user through power phase which traditionally were limiting to amputees and prosthesis wearers. These power phases can accommodate climbing stairs and slopes, lifting oneself from a chair or avoiding tripping in ordinary walking, all the while reducing metabolic exertion through ordinary activities. The infrastructure of the system also accommodates adaptive development of further more complex gait modes in the future due to the unique algorithm developed by Ӧssur engineers which accomplishes the challenge of walking with a missing leg from above or though the knee but in unique fashion, exploiting the fundamental characteristics of gait with an elegant algorithm, innovating where other players in the market have focussed on passive devices which still have their limitations. Whilst MMG technology can equally be applied to passive devices, the potential of the technology is exemplified in this current project due to the power to observe complex muscle activation signals from the user’ residuum, something that has so far only been accomplished with EMG systems. Even EMG has been slow to reach the market with Ӧssur now experimenting with mature trials of implanted EMG ‘pills’ which can overcome some of the limitations in surface EMG (sEMG) that our current project is overcoming through novel use of the MMG sensors. It is our hope that the project will be validated by seeing an amputee get up and walk using muscle activation as derived from MMG microphones alone. The benefits to the life of the user are difficult to quantify but would be life changing as we are redeploying neural connective links otherwise lost by trauma, one large step towards natural and intuitive daily function and comfort for many thousands of people whose lives have been adversely affected by a life changing event.

In solving the problem at hand we are exploiting modern developments in machine learning, systems that can allow a computer to learn in a fashion otherwise limited to cognitive function of the brain. The computer system embedded in the prosthetic knee will observe the user walking and infer from this how it will allow the user to control the knee themselves. The control could be through conscious or sub conscious control, one of many factors in development with this project. The combination of innovation in powered actuated knees and novel muscular control lend this project a weight that may help its propagation into the industry and the lives of those whose lives are less enabled by devastating life events. Even incremental developments in this field can have such amazing benefits to these people and drives us day to day to push the limits of what is capable with current technology.