R-HATS (RCO Human-Automation Teaming System) is a million NASA funded award in which HATS developed and tested technology and concepts which address and support interaction and task exchange between humans and automation, and extended traditional human-only crew resource management (CRM) to include ground and air-based automated agents with the humans (on-board/remote pilots and dispatchers). This CRM approach brings automation onto the tradition “team”, allowing human or automation to truly collaborate and complement each other’s strengths and weaknesses. We achieve this by using two innovations: (i) Mixed Initiative System design, in which both human and automated agents autonomously propose tasks and act independently; (ii) machine learning algorithms, which will allow the automation to improve over time as more data from R-HATS is used to train the system. These approaches are integral to the proposed capacity to tailor the automation to individual human team members and to allow a collaboration process closely mirroring the human teaming currently promoted by the tenets of CRM.
R-HATS was developed and integrated into NASA labs using an integrated product development team (IPT) with NASA as a customer and close collaborator. Our work plan included: (1) a development strategy based on conducting analyses of on-board automation/autonomy technologies, analyzing tasks in nominal and off-nominal conditions for representative use cases that involve different levels of degradation in the aircraft, environment, and the pilot, and conducting rapid, iterative human in the loop (HITL) simulations to identify choke points and high workload situations; (2) use of existing strategies to design technologies and interfaces based on an initial architectural design for R-HATS, notional mockup interfaces, incorporation of dynamic task allocation, a verbal interface and controlled naturalistic audio, an initial Pilot Health Monitoring System and a machine learning module; (3) iterative tests of R-HATS for nominal/off-nominal conditions using rigorous software/hardware development processes and design reviews with NASA; (4) integration of R-HATS into NASA facilities, including a shakedown simulation, with delivery of all source software/hardware and documentation; and (5) conducting real-time HITL simulations at NASA to study a NASA-chosen research question by evaluating a baseline condition with and without human automation teaming.