<![CDATA[As time progresses, technology also experiences rapid advancement. However, this advancement is not accompanied by an improvement in the interaction between computers and humans. This is certainly a significant impact on users of technology, particularly those with mobility disabilities. Smart wheelchair technology has already been widely used to assist people with disabilities, but the problem of interaction with computers still reduces user comfort. Some new types of interaction have been offered in previous research, including using voice, but this method is considered less effective due to the need for a minimally noisy environment. Another offered method of interaction is using eye gaze estimation using conventional algorithms or machine learning. For this, conventional algorithms are considered less effective due to their low adaptability with different users. CNN-based algorithms are chosen in this research because of their ability to extract features from images, allowing the algorithm to adapt to new data. The accuracy of the model in this research was 96% with a loss of 0.02 during the training phase. The system can run the algorithm in 0.16 seconds using CUDA acceleration. The system only uses 12 watts of electricity, making it possible to run the system using a battery. From the testing that was carried out and the results obtained, it can be concluded that the system runs well to estimate the direction of the user.]]>
Copyrights © 2022
