Brain-Robot Interface Research Group
Optimizing Convolutional Neural Networks to Control the Robotic Hand using Brain Signals
In recent years, machine learning and deep learning have been used in a variety of domains such as natural language processing and image recognition. In particular, the application of deep learning in robotics are mainly focused on image processing and object recognition. In our laboratory, we integrate Genetic Algorithm and Convolution Neural Networks to map brain signals to correct robot hand motion. The GA optimizes the hyperparameters of CNNs to improve the EEG classification rate. In addition, we build the model that is robust against data partitioning. Trained CNNs are implemented to control in real time the robotic hand using brain signals.