Brain-robot Communications in the Internet of Things

Abstract

Transformation of technologies to operate in Internet of Things (IoT) will increase their connectivity for sharing and collecting data with minimal human intervention, as well as to use advantages of cloud computing. The paper presents a novel approach for integrating EEG brain computer interface (BCI) with a humanoid robot for communications in the IoT. A transformation of both devices into IoT things allow seamless communication for data exchange with other IoT devices or services and more efficient computations in the web or cloud. Node-RED is the chosen programming tool providing a gateway to IoT, by which the EEG-based Emotiv EPOC+BCI device and the humanoid robot Pepper communicate. There is a library of input nodes in Node-RED for EPOC+ device, by which it interfaces services and devices in Internet, however such nodes do not exists for Pepper. Two approaches for transformation of the robot into IoT device have been designed and tested. In the first, MQTT protocol has been deployed on the robot side that can use local or remote MQTT broker. In the second approach, there is no need to install specific publish-subscribe client software for interactions with the broker. Instead, continuous python process in Node-RED accesses the Pepper internal memory for storing and retrieving key-value pairs. The devices share and collect data based on publish-subscribe model embedded in Node-RED. Original flows were designed and developed for BCI control of the robot in the IoT. The control is based on EEG data featuring, classifying and translating into machine commands that interface the robot APIs from Node-RED.