User-Independent Intent Recognition on a Powered Transfermoral Prosthesis
Abstract
This research will develop intuitive and smart intent recognition systems for powered prostheses to predict user intent to optimally supply power to the gait cycle during locomotion tasks. Intelligent intent recognition systems are needed for these prostheses to be clinically deployable. The primary scope of this project first involves developing and preparing a powered prosthesis complete with control technologies for clinical testing with patients with transfemoral amputation. We will collect data during walking which includes various speeds, stairs and ramps. We will compare the clinical effectiveness of different intent recognition systems on lower limb amputees using a powered prosthesis. This research will result in clinically meaningful parameters including the success rate, speed and energetic cost of the amputees performing a circuit of locomotion activities including level walking, stairs and ramps. Biomechanics of movement and energetic cost using the controllers will be quantified and compared to passive prosthesis ambulation. Results to date include the improved development of the mechanical, electrical, and control systems for the powered prosthesis. We have performed tuning and data collection on 9 subjects. This data has gone into the refinement of machine learning algorithms for seamlessly and continuously estimating walking speed and ground slope to improve community ambulation. Results show the relative ease of using our controllers as well as showing improved functionality in different ambulation compared to their conventional passive prostheses.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2019
- Accession Number
- AD1087123
Entities
People
- Aaron Young
- Krishan Bhakta
Organizations
- Georgia Tech