Related Experiment Video
Updated: Mar 6, 2026
11:16
Engineering Platform and Experimental Protocol for Design and Evaluation of a Neurally-controlled Powered Transfemoral Prosthesis
Published on: July 22, 2014
16.7K
Preliminary results for an adaptive pattern recognition system for novel users using a powered lower limb prosthesis.
Summary
This study introduces an adaptive pattern recognition system for powered prosthetic legs. The system learns user-specific data during ambulation, significantly reducing errors compared to non-adaptive systems.
Area of Science:
- Biomedical Engineering
- Rehabilitation Robotics
- Human-Computer Interaction
Background:
- Powered prosthetic legs enhance lower limb amputee mobility.
- Pattern recognition systems enable seamless transitions between locomotion modes.
- Current systems require extensive, burdensome training data collection.
Purpose of the Study:
- To develop an adaptive pattern recognition system for powered prostheses.
- To reduce the data acquisition burden for users.
- To enable automatic, real-time learning of subject-specific gait patterns.
Main Methods:
- An adaptive pattern recognition system was developed.
- The system learns automatically from subject-specific data during ambulation.
- Tested on two able-bodied subjects using powered knee-ankle prostheses as novel users.
Main Results:
- Initially high error rates decreased as the adaptive system incorporated subject-specific data.
- The adaptive system reduced errors by 32.9% [8.6%] compared to non-adaptive systems.
- Demonstrated effective real-time learning and adaptation.
Conclusions:
- Adaptive pattern recognition systems offer significant improvements over non-adaptive systems.
- This technology has the potential to enhance the usability and performance of powered prostheses.
- Reduces training burden, leading to more seamless prosthetic integration.
