Tools and methods for Human-Autonomy Teaming: Contributions to cognitive state monitoring and system adaptation
share
The Human-Autonomy Teaming paradigm (HAT) has recently emerged to model and design hybrid teams, where a human operator must cooperate with an artificial agent, able to independently evolve in dynamic and uncertain situations. An important challenge in HAT is to transform autonomous systems into better teammates, capable of joining humans in highly interdependent activities. The presented works explore two main avenues, supported by industrial collaborations (in the domain of transportation and industrial systems), academic partnerships (especially with South Australian universities), and with the supervision PhD students. The first axis deals with the monitoring of cognitive states, to equip the machine with an ability to detect when human face difficulties. To address this question, a global approach is proposed to classify operators mental workload from the fusion of multisourced physiological and behavioral data. The second axis focused on the mechanisms for adapting human-autonomy teaming, making machine more compatible with human. Two kinds of solution are explored. One focused on the offline enhancement of the know-how-to-cooperate of machines, with the aid of CWA method and MDE techniques. The other deals with online adaptation of human-machine cooperation, where autonomous system can be considered inside the team - as a teammate - or above the team-as a coach. Finally, new research directions are opened, supported by ongoing initiatives in France and abroad. These perspectives relate to the consolidation of a multilevel approach for cognitive state monitoring, the building of a transparent dialogue between human and autonomy, a deeper consideration of transitional and longitudinal situations in HAT, and the scale-up challenge of studying HAT with human teams.
READ FULL TEXT