The aim of this project is to provide user assistance in the multi-tasking example scenario based on the MATBII task space developed by NASA. The attentional workspace of the user is monitored through their interactions and eye movements by and the underlying agent architecture which in turn is able to deploy highlights and gaze contingent arrows to guide the user’s attention.
An implementation of MATBII in Python 3.6+ with additional user assistance features. ICU is implemented as an event system that is easy to interface with. Interface meta data and all internal events are exposed via a simple event-based python API. ICU includes support for various kinds of user input - mouse, keyboard, joy stick, eye tracking.
A full version of system implemented with simple agents controlling the display of additional highlights and arrows according to simple logic based on eye tracking, keyboard, mouse and joystick inputs.
This project is supported by a Human-Like Computing EPSRC Network+ kickstart grant
A demo of the ICU + Agents system with mouse based attention. Mouse movements are classified as gaze/succade and agents are tasked with guiding the users attention by analysising the ICU event stream in real time.
Warnings (red highlights) are shown by the agents according to a set of rules. In this example a warning is triggered for a particular component is in a undesirable state and the user hasn’t recently interacted with component or task.”