Design of a Dexterous Robotic Arm Manipulator Using Hybrid BCI

Abstract

The paper describes the design of a hybrid Brain Computer Interface (BCI) system that provides control commands to manipulate a robotic arm. The goal is to facilitate BCI controlled real-time robotic applications by using a semi-autonomous operation mode that accepts multiple commands. Simple tasks, like moving forward or turning, are executed based on a single BCI command while more complicated tasks, like grabbing or pushing an object are automated once the task is selected. The robotic arm vision system uses an Intel RealSense D435 camera for image and depth perception where a point cloud generates an interface for the user to select an object. The user selects an object to manipulate, which identifies the goal position and location. After the location and object is determined, the software interface moves the robotic arm to have the selected object in the robotic arms’ workspace. The current robotic arm design utilizes an open bionics Brunel dexterous hand as the end effector which allows for human-like hand actions. A simulation platform is developed to verify the effect of the entire system of a dexterous robotic arm on a mobile platform. The system design and results using a hybrid BCI system is demonstrated.