Hand Gesture Recognition for Assisting the Elderly

With the increasing elderly population who cannot take care of themselves, the need for human-machine interaction that does not require tedious physical effort is growing. Hand gestures have the potential to provide a non-intrusive and intuitive means of interaction for Human Machine Interface (HMI). The project aims to develop a hand gesture recognition system for the NAO robot, which can respond to gestures provided by elderly people. To achieve this goal, several machine learning and deep learning models, including RF, SVM, and CNN, were compared to find the best-fitting model for hand-gesture recognition. The dataset used for training these models was obtained from Kaggle, and Python was used as the primary language for model execution. The performance of these models was evaluated using specific performance metrics with which a most optimal model is then chosen for the NAO bot. This approach has the potential to revolutionize the elderly care industry, enabling the elderly to interact with humanoid robots effortlessly, and thereby improving their quality of life.