Brain Teleoperation of a Mobile Robot Using Deep Learning Technique

This paper proposes a brain-teleoperation control strategy that combines the deep learning technique (DLT) to realize the control and navigation of a mobile robot in unknown environments. The support vector machine (SVM) algorithm is utilized to recognize the human electroencephalograph (EEG) signals in the brain-computer interface (BCI) system which is based on steady state visually evoked potentials (SSVEP). In this way the intentions of human can be distinguished and control commands are generated for mobile robot. The DLT is used to recognize the type of environmental obstacles and environmental features by analysing the images that describe the environment. And then according to the classification of obstacle, various potential fields are built for the specific obstacles. By utilizing bottles as the features of environment, a whole map of the surroundings can be built through a sequential simultaneous localization and mapping (SLAM) algorithm. The main contribution of this paper is that the relationship between the potential field strength and classification of EEG signals is built up through the combination of multiple artificial potential fields with the brain signals, which produces the motion commands and designs a trajectory free of obstacles in un-structure environments. Three volunteer subjects are invited to test the entire system, and all operators can successfully complete experiments of manipulating the robot in corridor environments.