1. Computational Evaluation of the Allocation of Authority and Responsibility in NextGen Concepts of Operation Thesis Proposal for MS in Aerospace Engineering By Raunak P. Bhattacharyya Georgia Institute of Technology 09/10/15

2. Can We Define NextGen as Technologies? 2 By itself, better technology will not result in transformative developments. Need novel concepts of operation 1 1 Department of Transportation Office of Inspector General Audit Report, 2014

3. Concept of Operation Specification of how an operation is to be carried out Using available resources Actors who will execute actions Air traffic concept of operation Actions to be carried out to reach desired outcome By whom Conditions where each action needs to be carried out Criteria for successful completion of each action 3

4. Allocation of Actions to Agents Autonomy: Actions that an agent is capable of performing independently Authority: Which actions an agent is asked to perform Responsibility: Which outcomes an agent will be held accountable for 4 1 Pritchett, Kim and Feigh, JCEDM, 2014 Authority-Responsibility mismatch creates monitoring reqmts 1 Allocation is the process of defining which agent has authority for actions and which agent has responsibility for the outcome

5. Latency Delay in communication Stale information Cascade effect throughout the system Impact of latency depends on allocation Information exchange between agents 5 Latency = 10 sec

6. Modeling Concepts of Operation In model Physics of flight (6 DOF) Actions Resources Agents During runtime Allocation Latency Tasks done by agents Information transferred between agents Global properties of system of aircraft 6

7. Emergence Allocation and latency are a part of a mid-level model of concept of operations Effect of interactions at one level transcend/percolate to higher/lower levels 1 Two levels Emergence ‘up’ Emergence ‘down’ 7 Mid-Level Model of Concept of Operations (Physics, Actions, Allocation) Mid-Level Model of Concept of Operations (Physics, Actions, Allocation) Spacing achieved Taskload Information Transfer Requirements Emergence Up Emergence Down 1 Leveson, Safeware, 1995

8. Designing Concepts of Operation Current Day Lack systematic, early-in-design method to analyze allocations and communication constraints Rely on subject matter experts Testing methods rely on HITL Proposed 1,2 Model concept of operation early-in-design Analyze impact of allocation and communication constraints through computational simulation Identify key issues through capturing emergent behaviors Use results to guide procedures and technology design 8 1 IJtsma, Bhattacharyya & Pritchett, ISAP 2015; 2 IJstma,Bhattacharyya & Pritchett, ATM 2015

9. Objectives 1.Investigate the impact of allocation of authority and responsibility on agent taskload and information transfer requirements with no latency;  Emergence down 2.Investigate the impact of latency on system performance regardless of allocation of authority and responsibility between agents; and  Emergence up 3.Investigate the combined impact of allocation and latency on both high- level system performance and on agent level measures of taskload and information transfer requirements. 9

10. Case Study: Merging and Spacing 10

11. Modeling Concept: Actions and Functions 11 Functional blocksActionsAction description Vertical profile control Calculate distance to runwayCalculate the remaining distance to the runway. Start descentWhen clearance is given, start the 2 degree descent Intercept GSIntercept glideslope signal and initiate 3 degree glideslope descent Land aircraftLand the aircraft (remove aircraft from simulation) Vertical profile management Clear for descentGive clearance to start the descent at TOD. Clear for final approachGive clearance to start final approach and intercept ILS signal. Lateral control Direct to waypointSet the heading based on the target waypoint. Calculate distance to waypoint Calculate the distance to the next waypoint. Lateral profile management Manage waypointSet the target waypoint. Execute path stretchExecute the maneuver by entering it into the autoflight system.

12. Research Plan 12 AllocationAllocation Taskload Information requirements Emergence Dn Stage 1 Latency Spacing Emergence Up Stage 2 Stage 31 Sim Engine

13. Stage 1: Scenario Description Aircraft need to maintain 60 sec time spacing PD controller which controls commanded airspeed Simple controller to isolate problems with concept More detailed controller can be used later in design Authority and Responsibility are allocated to Air and Ground agents Agents are perfect Actions are instantaneous and perfectly executed 13 This work was done in collaboration with Martijn Ijtsma, TU Deflt

14. Functions Authority allocations (AA) Responsibility allocations (RA) 1234512345 Vertical controlGAAAAGAAAA A/C config mgmtGAAAAGAAAA Lateral controlGAAAA GAAAA Speed controlGGAAA GGAAA Lateral profileGGGAAGGGAA Vertical profileGGGGAGGGGA Speed mgmtGGGGAGGGGA Non-nominal situation mgmt GGGGAGGGGA Allocation Matrix 14

15. Incoherent Allocation 15 Functional blocksActionsAuthorityResponsibility Vertical profile control Calculate distance to runwayGA Start descentAG Intercept GSAG Land aircraftAG Aircraft configuration management Set flaps and speedbrakesAG Deploy gearAG Lateral control Direct to waypointGA Calculate distance to waypointAG Speed controlSet airspeedGA Lateral profile management Manage waypointAG Execute path stretchAG Vertical profile management Clear for descentAG Clear for final approachGA

16. Time Trace of Actions: AA2 RA4 16 Total 26 action traces: 25 authority and responsibility allocations, 1 incoherent allocation

17. Taskload 17 Authority-responsibility mismatch creates additional work

18. Information Transfer Requirements 18 Incoherence increases information transfer reqmts drastically

19. Key Findings Coherence can be used as a guiding principle in the design of concept of operations  Higher incoherence demands higher information transfer requirements Authority-Responsibility mismatches need to be reduced and where inevitable, need to be identified early in design  Mismatches create additional taskload in terms of monitoring that are not explicit in the required work to be done in the concept 19

20. Checkpoint  Investigate the impact of allocation of authority and responsibility on agent taskload and information transfer requirements with no latency  Modeled spacing and merging operations for 3 aircraft STAR  Analyzed 26 allocation scenarios including an incoherent allocation scheme  Emergent down metrics taskload and information transfer requirements were analyzed  Investigate the impact of latency on system performance regardless of allocation of authority and responsibility between agents  Investigate the combined impact of allocation and latency on both high- level system performance and on agent level measures of taskload and information transfer 20

21. Stage 2: Latency Test to Failure Delay in communicating commanded airspeed to maintain spacing Analytical latency margin Computational latency experiment: test to failure 21 Block diagram of spacing controller

22. Stage 3: Scenarios and Test Conditions AllocationNo latencyBelow instabilityAt instability Authority1 - Responsibility1 … Authority5 – Responsibility5 Incoherent 22 Stream of 10 aircraft

23. Timeline Research TaskTime Required Estimated Completion Date Latency: Time delay margin analysis1 weekSep 22 Latency: Sensitivity analysis, test to failure2 weeksOct 6 Analysis and write up of results3 weeksOct 27 Latency and allocation simultaneously3 weeksNov 17 Analysis and write up of results4 weeksDec 15 Assemble thesis2 weeksDec 29 23

24. Anticipated Contributions Concept can be tested early in design without needing technology prototypes Analyze impact of authority responsibility mismatches Pinpoint when latency causes instability Predict emergent behaviors both at local and at global level, and also relate them to linear modeling Analyze relative costs and benefits of proposed allocations 24

25. Acknowledgements Dr. Amy Pritchett Dr. JP Clarke and Dr. Magnus Egerstedt Martijn IJtsma, TU Delft Cognitive Engineering Center NASA Aviation Safety Program with Guillame Brat of NASA Ames Research Center as technical monitor 25