Modeling the impact of increasing autonomy on core cognitive abilities in unmanned system operation
Abstract
Funding is provided for: Task 1: Develop a simulation test bed that simulates handheld control of a small tactical UAVthat reflects both current capabilities, as well as those of increased autonomy.~ A preliminary version was designed using USARsim to simulate indoor control ofquadrotors (Jackson, 2012) but this will be modified to match the outdoor tactical userepresentation of USMC expeditionary tasks.Task 2: Develop a simulation test bed that simulates the ability of one operator to controlmultiple UAVs, reflecting current operations as well as another version that simulates increasingautonomy.~ The simulation will leverage the RESCHU simulation, developed under a previousScience of Autonomy effort (Nehme, 2009) and now used across the Navy in researchorganizations like NRL, NPS and SPAWAR as well as universities worldwide.Task 3: Develop a general system dynamics model of the relationship between skill-, rule-,knowledge-, and expert-based (SRKE) behaviors.~ Task 3a: Determine a principled measurement approach of SRKE behaviors and theirrelated performance measures. While a formal taxonomy for measurement of human andautonomous system performance has been proposed and used throughout DoD researchefforts (Donmez, Pina, & Cummings, 2008), this taxonomy did not attempt to developany metric classes for these behaviors, which this proposed effort will develop (and thusextended the value of the taxonomy).~ Task 3b: Tailor the generalized system dynamics model to the small, tactical UAVdomain and make predictions with the model as to how SRKE measures and theirinterdependencies will be affected by increasing system autonomy.~ Task 3c: Tailor the generalized system dynamics model to the large, more strategic-levelmultiple UAV domain and make predictions with the model as to how SRKE measuresand their interdependencies will be affected by increasing system autonomy.~ Initial validation of these models will occur with pre-existing data sets used from relatedresearch efforts.Task 4: Conduct human-in-the-loop experiments with both simulations to assess and validate themodel. And in each simulation environment there will be at least two different autonomy levels(current and future), with possibly more.Task 5: Update the model according to the results:~ Task 5a; Use the model to analyze questions such as, ~When should functions not beautomated to support skill and/or rule development?~ and ~Are there functions thatshould be automated as they have no measurable impact on performance?~~ Task 5b: Disseminate the results to various Naval and other interested DoDorganizations.Option Period (2 Years)Task 6: Extend this modeling method and experimentation to other domains including groundrobots that typical of those currently used and expected-to-be used in Naval and Marine Corpssettings. We would then compare and contrast the finding with air domain to determine whatinterdependencies are domain specific and those that may cut across different robotic systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141612609
Entities
People
- Missy Cummings
Organizations
- Duke University
- Office of Naval Research
- United States Navy