Related Experiment Video
Updated: Dec 30, 2025
11:16
Engineering Platform and Experimental Protocol for Design and Evaluation of a Neurally-controlled Powered Transfemoral Prosthesis
Published on: July 22, 2014
16.6K
Obstacle Recognition using Computer Vision and Convolutional Neural Networks for Powered Prosthetic Leg Applications
Summary
This study developed a computer vision system using Convolutional Neural Networks (CNNs) to help powered prosthetic legs identify obstacles like stairs and doors, achieving 90% accuracy.
Area of Science:
- Robotics
- Computer Vision
- Machine Learning
Background:
- Powered prosthetic legs aim to enhance mobility for amputees.
- Current prosthetics lack real-time environmental awareness.
- Anticipating walking obstacles is crucial for user safety and seamless navigation.
Purpose of the Study:
- To develop a computer vision system for powered prosthetic legs.
- To enable prosthetics to identify common walking obstacles like stairs and doors.
- To improve prosthetic leg synchronicity with user intent through environmental recognition.
Main Methods:
- Combined computer vision with Convolutional Neural Networks (CNNs) for obstacle identification.
- Utilized a compact CNN architecture for optimized real-time image processing.
- Developed and tested a wearable prototype with a camera and single-board computer.
- Collected and labeled video data from able-bodied users to train the CNN model.
Main Results:
- The system achieved approximately 90% accuracy in recognizing obstacles.
- Successful identification of stairs and doors in diverse indoor and outdoor environments.
- Demonstrated the practicality of a wearable system for powered prosthetic leg applications.
Conclusions:
- Computer vision and machine learning show significant potential for powered prosthetic leg advancement.
- The developed system offers a viable solution for real-time obstacle detection.
- This technology can enhance the autonomy and safety of prosthetic leg users.
