Real-time Prediction of Human Error
Human error in industrial and clinical applications can be associated with high risk and cost. In this work, we developed an application for Microsoft HoloLens 2 that can predict a user’s mistakes before they are executed.
Our system utilizes a fundamental human behavior, namely hand-eye coordination, in which the eyes often look ahead to plan the motor systems movement. By tracking eye-gaze and hand movement, our system can anticipate what a user will do next and, in case of a mistake, warn the user in advance.
While there are direct implications of our work for industrial applications such as maintenance, we also see great potential in tracking hand-eye coordination in surgical procedures to provide task- and skill-dependent assistance.
Published in IEEE ISMAR: external page https://doi.org/10.1109/ISMAR52148.2021.00031
