ABSTRACT
The problems associated with sensors and cognitive state gauges seem to grow exponentially as we seek to transform them from a laboratory phenomenon into a deployable system. Solutions to the problems dictate the types of mitigations that can be employed and the operator interface software architecture. Some of the problem areas include:
Sensor Noise: Sensors can have good sensitivity and accuracy, and still be too noisy to be used in an applied system. For example, a sensor may correctly show high workload 90% of the time, and yet still oscillate between high (where it stays 90% of the time) and low often enough to confuse an augmentation manager trying to select the correct mitigation.
