Experimental characterisation of eye-tracking sensors for adaptive human-machine systems

Adaptive Human-Machine Interfaces and Interactions (HMI ² ) are closed-loop cyber-physical systems comprising a network of sensor for measuring human, environmental and mission parameters, in conjunction with suitable software for adapting the HMI ² (command, control and display functions) in response to these real-time measurements. Cognitive HMI ² are a particular subclass of these systems, which support dynamic HMI ² adaptations based on the user's cognitive state. These states are estimated in real-time based on various biometric parameters from gaze, cardiorespiratory and brain signals, which can be fused using suitable models. However, the accuracy and precision of biometric measurements are affected by a variety of environmental factors and therefore need to be accurately characterised prior to operational use. This paper describes the characterisation activities for two types of eye tracking devices available in the Aerospace Intelligent and Autonomous Systems (AIAS) laboratory of RMIT University, being used to support the development of cognitive human-machine systems. To classify the user's cognitive states, eye-tracking features are processed by a machine learning classifier based on fuzzy logic. This paper describes how the uncertainty associated with the classified outputs is quantified by propagating the uncertainty of the input features, which was characterised experimentally, through the classifier. This process is of growing relevance because machine learning classifiers are of increasingly common use, therefore it is discussed in detail in the paper.