Deep learning neural networks for NAO robot control

• Main Author: Tee, Enid Mun Xin
• Other Authors: Song Qing
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/78219
• Institution: Nanyang Technological University

Deep Learning is one of the solutions to the future of technology. It is a subset of machine learning. With deep learning, it is possible to ‘learn’ and make ‘informed choices’ based on the data it has analysed. With the addition of deep learning to eye trackers, it is possible for the device to understand the user’s habits and improve the prediction of the output. In this project, deep learning algorithm is applied to the NTU’s patented eye tracking technology to help improve the stability and the predicted outcome of eye gazes. It will be able to classify certain eye gestures, which allows the user to activate certain commands. The software component will be developed using Microsoft Visual Studio (MVS) as well as Open Source Computer Vision Library (OpenCV). The language used will be C++. For deep learning, a Recurrent Neural Network (RNN) will be developed using Matrix Laboratory (Matlab) as it has built in functions that can support deep learning training easily.