The idea of using a powered wheelchair, for people with mobility limitation and the elderly has been around for quite a while. Most of these wheelchairs require the use of upper limbs to control them. On the contrary, this project aims to help quadriplegic individuals to use their wheelchair with minimum human assistance. It involves the use of Bio-signals mainly EMG EOG and EEG to control the intelligent wheelchair using Artificial Neural Network and Sensor Fusion technology. The setup can also be use for below the neck paralyzed or elderly people with less upper arm strength.
It’s a new approach towards wheelchair control which is non-invasive, discrete and functional. This document gives details of the human-machine interface, the technical equipment, functionality, evaluation and implementation of the system.
Table of Contents
1 INTRODUCTION
1.1 History
1.2 Types of wheelchairs
Manual wheelchairs
Electric-powered wheelchairs
Limitation of the electric chairs
Smart or Intelligent wheelchair
1.3 Current research
2 GOALS AND OBJECTIVE
2.1 Project description
2.2 Hardware
Data acquisition box
Wheelchair
2.3 Software
Software description
Data collection
Pre-processing and data segmentation
Feature extraction
Classification
Algorithm
Training and Testing
Normalizing
Decision process
User interface
3 EVALUATION
Component testing
Sub-system testing
System testing
4 FUTURE DEVELOPMENT
4.1 Challenges faced and recommendations for future work
5 PROJECT PLAN
Work breakdown structure
Gantt chart
5.1 Conclusion
Project Goals and Themes
This project aims to design and implement a hybrid control algorithm for an intelligent wheelchair, specifically targeting the needs of quadriplegic individuals. By utilizing bio-signals (EEG, EOG, and EMG) captured via a headband, the system enables users to navigate a wheelchair with minimal human assistance, significantly improving their independence and social mobility.
- Integration of hybrid bio-signal processing (EEG, EOG, EMG)
- Development of pattern recognition using Artificial Neural Networks (ANN)
- Design of a human-machine interface for wheelchair control
- Real-time signal acquisition and noise-reduction strategies
- System evaluation through component, sub-system, and full system testing
Excerpt from the Book
Feature extraction
Feature extraction stage is a challenging stage. The features extracted from the signal should be sufficient to represent the movement which triggered it. The data obtained from data segmentation stage is used to narrow the search to select the features at the precise moment the signals are triggered. The amplitude or phase thresholds of the signals are further reduced to extract features from the data samples. Once the data range is available three features are extracted from the samples. The features extracted are Absolute Mean Value (AMV), Root Mean Square value (RMS) and Average Crossing value (AC). The extracted features are then saved in a file for training the artificial neural network.
In the project both quantitative and qualitative methods were implemented. During data collection the theoretical knowledge of Bio-signals was applied for selecting the signals. The qualitative approach was used to find the thresholds for each pattern, for appropriate control of the wheelchair. For FSC recognition a gradient function from the EMG signal is calculated which gives the deviation value at the nth sampling. N is the total number of samples.
Summary of Chapters
1 INTRODUCTION: Outlines the necessity of assistive technologies for the disabled and elderly, introducing the role of bio-signals in human-machine interaction.
2 GOALS AND OBJECTIVE: Defines the project's purpose, detailing the system architecture, hardware components, and the software methodology for signal processing and ANN classification.
3 EVALUATION: Discusses the three-tier testing process, covering individual component performance, sub-system functionality, and overall system reliability in real-time scenarios.
4 FUTURE DEVELOPMENT: Addresses observed limitations in signal precision and recommends enhancements, such as adding more sensor channels and utilizing advanced pattern matching algorithms.
5 PROJECT PLAN: Details the time management, work breakdown structure, and scheduling strategies employed to complete the project phases.
Keywords
Human Machine Interaction, Intelligent wheelchair, EEG, EMG, EOG, Artificial Neural Network, ANN, Bio-signals, Quadriplegic, Pattern Recognition, Signal Processing, Assistive Technology, Cyberlink, Control Algorithm, Rehabilitation.
Frequently Asked Questions
What is the primary focus of this project?
The project focuses on developing an intelligent control system for a powered wheelchair that allows individuals with mobility limitations, specifically quadriplegic patients, to navigate independently using bio-signals instead of manual controls.
Which specific bio-signals are utilized for control?
The system records and processes Electromyogram (EMG) signals from facial muscles, Electrooculogram (EOG) signals from eye movements, and Electroencephalogram (EEG) signals from the brain.
What is the core objective of the research?
The main goal is to create a robust and reliable hybrid bio-signal algorithm that translates specific user inputs—such as forehead movements and eye glances—into distinct navigation commands for the wheelchair.
How is the signal classification handled?
The system employs an Artificial Neural Network (ANN) as a classifier. It is trained to recognize specific movement patterns from the pre-processed bio-signal data and map these to motor commands.
What does the main body of the work cover?
The main body details the hardware setup (including the headband sensors and the wheelchair's internal processor), the software logic for data collection and segmentation, feature extraction, ANN training/testing, and the final decision-making process.
Which keywords best describe the work?
Key terms include Human-Machine Interaction (HMI), Intelligent Wheelchair, Bio-signals (EEG, EMG, EOG), Artificial Neural Networks (ANN), and assistive robotics.
Why were forehead-based bio-signals preferred for this project?
The forehead provides a non-invasive, accessible, and less intrusive location for capturing useful signals compared to traditional electrode caps, making it more socially acceptable and easier for paralyzed individuals to use.
What was a major challenge mentioned in the implementation?
A significant challenge was the real-time extraction of clean control signals from noisy face movements, requiring a trade-off between algorithmic complexity and system performance during execution.
What role does the fatigue indicator serve?
The fatigue indicator monitors signal strength and usage duration to prevent false command triggering, ensuring safe operation if the user is exhausted or if sensor contact quality degrades over time.
- Quote paper
- Pankaj Kadam (Author), 2010, Powered Wheelchair Controller Using Hybrid Bio-Signals, Munich, GRIN Verlag, https://www.grin.com/document/210927
