Abstract

In this paper, probabilistic-based workspace scan modes of a robot manipulator are presented. The scan modes are governed by a Brain Computer Interface (BCI) based on Event Related Potentials (Synchronization and Desynchronization events). The user is capable to select a specific position at the robot’s workspace, which should be reached by the manipulator. The robot workspace is divided into cells. Each cell has a probability value associated to it. Once the robot reaches a cell, its probability value is updated. The mode the scans are made is determined by the probability of all cells at the workspace. The updating process is governed by a recursive Bayes algorithm. A performance comparison between a sequential scan mode and the ones proposed here is presented. Mathematical derivations and experimental results are also shown in this paper.