الفهرس 
CHAPTER 1: INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 83 |  |  |
| | | from | 83 | | |
Abstract
Problem Statement:
Individuals with disabilities face numerous challenges in their daily lives, and even simple tasks like turning on a light bulb or seeking assistance from family members can be difficult for them. According to the World Health Organization, approximately 15% of the world’s population has disabilities. These individuals encounter challenges when dealing with their surrounding electrical devices.
The scientific research community has shown interest in addressing this problem, and researchers have proposed various solutions to help people with disabilities live more easily using technology. Scientific research has focused on assisting them in controlling their surrounding devices, such as lighting, fans, and other electrical appliances.
1. Researchers have utilized Infrared (IR) technology to control devices remotely. However, this solution posed a challenge as it required direct line-of-sight communication with the devices, and not all devices support this feature.
2. Another group of researchers explored using voice commands to control devices, but it was limited to using fixed or template commands such as ”turn on the light” or ”turn off the air conditioner.” Additionally, voice control often required an internet connection.
Objectives:
1. Develop an intelligent personal assistant that can be interacted with through voice commands.
2. Implement Natural Language Processing (NLP) to enable the model to understand indirect commands.
3. Create a practical application on Raspberry Pi and build a miniature model to showcase the results.
4. Establish a dataset for training the system.
5. Extract and measure the accuracy of the model’s results.
Methodology:
1. The system begins its operation by receiving voice commands from the user in an organized or unorganized format and converts them into text.
2. Voice is converted into text using the Google Speech-to-Text API if an internet connection is available, or the system uses the Python library ”pocketsphinx” for offline operation.
3. Natural Language Processing (NLP) technology divides the commands into separate words to identify the device to be controlled.
4. The system determines whether the text is a question or an executive command.
5. A logistic regression algorithm classifies the remaining words into ”turn on” or ”turn off” commands to operate the device.
6. The system sends the command to the Raspberry Pi for device execution.
7. The system reads feedback from the hardware and informs the user of the command’s success or any issues that occurred during execution.
8. The system returns to a standby mode to await the next command.
Results:
1. The proposed system was implemented using various algorithms such as logistic regression, Naïve Bayes, and support vector machines. It was trained on two datasets, one of which contained 3,000 regular, negative, and unstructured commands.
2. Simulation results indicate that the proposed system can accurately recognize 90% of device names for all commands.
3. The system correctly classifies nearly 100% of commands as positive or negative.
4. Command execution takes approximately 30 seconds, demonstrating the system’s potential to significantly improve the quality of life for individuals with disabilities by providing a more efficient and user-friendly way to control their household devices.
Recommendations:
Future work can focus on improving the system’s accuracy and expanding its capabilities to control more devices. Additionally, efforts should be directed toward enhancing offline functionality and improving the accuracy of text input to the system. In general, the proposed system has the potential to have a significant impact on the lives of individuals with disabilities and contribute to creating a more inclusive society. Future researchers can build upon this work to further benefit the community, continuing the scientific research journey from where others left off.