Abstract
The rapid advancement and increasing deployment of autonomous systems across various operational domains present new challenges for operator training and competency development in operators. As these systems become more intelligent and integral to mission success, there is a growing need for training that effectively prepares operators to collaborate with them. However, developing the necessary curriculums can be time-consuming, requiring instructors to align competency requirements with different system complexities and potential scenarios the operator may experience.
This paper presents a methodology for algorithmic curriculum development driven by the sources of complexity for a given system. These complexity sources are described by defining context scenarios relevant to the human-machine team operation. Our approach identifies complexity factors within the system that influence cognitive load, then uses them to generate an initial training curriculum that builds up complexity in a simulated environment to develop operator capabilities.
The methodology incorporates physiological biometric measures of cognitive load and self reporting of task complexity throughout the training process. By leveraging this data, alongside the operators' performance, the curriculum adjusts environmental complexity in the simulation between scenarios to maintain an optimal cognitive load, promoting efficient skill acquisition and competency development.
This adaptive curriculum approach aims to ensure that trainees are neither cognitively underloaded nor overloaded to maintain an appropriate pace of personalised training for the individual. The paper discusses the application of this methodology within a multi-agent human-machine teaming framework, demonstrating how adaptive complexity management can be used as a basis for curriculum development and training within simulated environments.