Human Factors. Principal Investigators: Nadine Sarter Christopher Wickens. Beth Schroeder Scott McCray. Smart Icing Systems Review, May 28,

Transcription

1 Human Factors Principal Investigators: Nadine Sarter Christopher Wickens Graduate Students: John McGuirl Beth Schroeder Scott McCray 5-1

2 SMART ICING SYSTEMS Research Organization Core Technologies Aerodynamics and Propulsion Flight Mechanics Controls and Sensor Integration Human Factors/ Cognitive Engineering Aircraft Icing Technology IMS Functions Characterize Icing Effects Operate and Monitor IPS Inform and Advise Pilots Envelope Protection Adaptive Control System Research/Integration Flight Simulation Flight Test 5-2

3 Cognitive Engineering Goal: Objectives: Approach: Improve the safety of flight in icing conditions. Develop smart system to improve ice tolerance. Design use-centered interface that a) informs pilots about presence/changes and performance effects of icing conditions b) communicates IMS/IPS status/activities/limitations to crew in timely and effective manner c) provides pilots with advisories for handling inflight icing encounters safely Identify pilots information requirements Develop candidates for human-centered cockpit interface Evaluate effectiveness and robustness of candidates in simulator studies 5-3

4 Information Requirements What To Present A Survey of Line Pilots Sent to 6,400 pilots (386 responses) from 9 regional carriers in collaboration with ALPA Ratings/explanation of importance of information on 3 areas: Characteristics of icing Aircraft configuration and performance IMS/IPS status/activities Description/Explanation of Critical Incidents 5-4

5 Information Requirements What To Present Reported Icing Monitoring Cues Visual check for ice accretion 93% Expectation-Driven Monitoring Loss of airspeed 69% Data-Driven Monitoring Need for additional power to maintain altitude/ climb Unusual trim adjustments by auto-pilot 20% 9% 5-5

6 Icing Encounter Detection IMS Functions Attention Capture/Guidance Diagnosis Monitoring Action Selection Status Display Trend Display Command Display Execution of Action Envelope Protection; Flight Control Adaptation 5-6

7 Attention Capture and Guidance Ambient strip captures attention Connected outline guides attention 5-7

8 Icing Encounter Detection Possible IMS Functions Attention Capture/Guidance Diagnosis Monitoring Action Selection Status Display Trend Display Command Display Execution of Action Envelope Protection; Flight Control Adaptation 5-8

9 Multimodal Information Presentation Visual and Tactile Cues For Presenting Icing-Related Information Modern flight decks impose considerable demands on visual and auditory channels Tactile channel is underutilized although powerful means of capturing attention and useful for providing some diagnostic information As more systems/data are added, multimodal information presentation becomes more important to avoid resource competition (Wickens Multiple Resource Theory) 5-9

10 Visual Conditions Visual additive Visual substitutive light moderate severe none 5-10

11 Multimodal Information Presentation Tactile Condition 2 vibrotactors on inside of forearm Cues presented sequentially (wing tail) and cycled for 5 seconds. Wing Tail 4 inches 5-11

12 Multimodal Information Presentation Detection of Icing Cues % Visual Cues Tactile Cues Tactile group performed as well as the two visual groups 5-12

13 Multimodal Information Presentation Secondary Visual Task Performance % VA VS TA VA VS TA Not concurrent Concurrent Tactile cues afforded better divided attention 5-13

14 Multimodal Information Presentation Summary 5-14

15 The IMS As A Decision Support System Icing Encounter Detection IMS Functions Attention Capture/Guidance Diagnosis Monitoring Action Selection Status Display Trend Display Command Display Execution of Action Envelope Protection; Flight Control Adaptation 5-15

16 Status and Command Displays A Simulator Study Comparing The Effectiveness of Status and Command Displays - Participants: 27 instructor pilots - Flight experience: average: 777 (827) hrs range: 200-4,600 hrs - 3 conditions: - baseline (no decision aid) - status display - command display - Medium-fidelity simulation of twin-engine aircraft 5-16

17 Status and Command Displays The Status Display The Command Display 5-17

18 Status and Command Displays Stall frequency as function of display condition and accuracy of IMS information % Stall Baseline Acc Stat Acc Com Inacc St Inacc Co Display Condition X Accuracy 5-18

19 Status and Command Displays Summary Status display appears to be preferable. equally beneficial with accurate information less vulnerable to effects of inaccurate information than command fewer recovery errors Still need better support for trust calibration as well as long-term planning and decision-making 5-19

20 Supporting Trust Calibration Use of automated systems, such as decision aids, has been linked to several factors including: - users confidence in performing the task - task complexity - risk involved in task - perceived and actual reliability of the automation Trust calibration refers to how closely perceived reliability matches actual reliability 5-20

21 Supporting Trust Calibration Participants: Flight experience: 30 U of I instructor pilots Average: 825 hrs ( hrs) Between-subjects variable: - reliability information (static vs. dynamic) Within-subjects variables: - type of DSS (command vs. status) - accuracy of decision aid (correct vs. incorrect) - familiarity with situation (wing vs. tail icing) - taskload (cruise vs. ILS approach 5-21

22 Supporting Trust Calibration Cockpit Display 5-22

23 Supporting Trust Calibration Confidence Display Provided a 5-minute history of reliability Y- axis values omitted to avoid fixation on a particular value Reliability was high for the first minute of each trial High Variable Low 5-23

24 Supporting Trust Calibration Stall frequency as a function of confidence information and decision aid accuracy % Stall Static Updated 0 Accurate Inaccurate 5-24

25 Supporting Trust Calibration Stall frequency as a function of decision aid type and decision aid accuracy % Stall (Schroeder, 2000) Status Command Accurate Inaccurate 5-25

26 Supporting Trust Calibration Pilot compliance with decision aid vs. DSS accuracy % Compliance Static Dynamic (high) Reliability averages to 70% overall Dynamic (variable) Dynamic (low) Reliability Compliance Rate 5-26

27 Supporting Trust Calibration Reversal of compliance as a function of reliability information display % trials with reversal Static Dynamic (high) 26 Dynamic (variable) 16 Dynamic (low) 5-27

28 Supporting Trust Calibration Detection of navigation-aid failure as a function of reliability information type % Failure Detection Static 20 Dynamic 5-28

29 Supporting Trust Calibration Summary Providing system reliability feedback afforded better trust calibration, resulting in less over-reliance and fewer stall events Also appears to have reduced automation bias, allowing for more flexible, adaptive responses for error recovery Given the added information, command display may be more desirable Further work is needed to explore situations which contain - less predictable reliability feedback - larger number of possible diagnoses 5-29

30 Overall Design Concept Sample Sequence of Possible Icing Encounter and Associated IMS Indications 5-30

31 Overall Design Concept 5-31

32 Overall Design Concept Attention Capture and Guidance Status Display IMS/IPS and System Confidence 5-32

33 Overall Design Concept 5-33

34 Envelope Protection Overall Design Concept 5-34

35 Overall Design Concept Visual Indications Should Be Combined With Auditory and Tactile Cues 5-35