Implications of automation and collaboration in cockpit operations: the road towards reduced crew operations

Transcription

1 Implications of automation and collaboration in cockpit operations: the road towards reduced crew operations Philippe Palanque - Professor University Toulouse III, France - palanque@irit.fr

2 The issue of function allocation Usability Body of Knowledge : Function allocation (also known as task allocation) is a classic human factors method for deciding whether a particular function will be accomplished by a person, technology (hardware or software) or some mix of person and technology. To do this, the investigator considers error rates, fatigue, costs, hazards, technological feasibility, human values, ethical issues, and the desire of people to perform the function. Jordan, N. (1963). Allocation of functions between man and machines in automated systems. Journal of Applied Psychology, 47,

3 The issue of function allocation Usability Body of Knowledge : Function allocation (also known as task allocation) is a classic human factors method for deciding whether a particular function will be accomplished by a person, technology (hardware or software) or some mix of person and technology. To do this, the investigator considers error rates, fatigue, costs, hazards, technological feasibility, human values, ethical issues, and the desire of people to perform the function. Jordan, N. (1963). Allocation of functions between man and machines in automated systems. Journal of Applied Psychology, 47,

4 The issue of function allocation Usability Body of Knowledge Function allocation (also known as task allocation) is Function a classic human = task factors method for deciding whether a particular function will be Entity accomplished = human/machine by a person, technology able to (hardware or execute software) a or task some mix of person and Execution technology. = input-output To do this, the / investigator analysis considers error rates, fatigue, costs, hazards, technological / edcision feasibility, / execution human values, ethical issues, and the desire of people to perform the function. Jordan, N. (1963). Allocation of functions between man and machines in automated systems. Journal of Applied Psychology, 47,

5 However There are many ways to allocate functions Allocate to systems (automation) Allocate to operators (collaboration) Allocate to both operators and systems (groupware) Issues to be addressed Satic versus dynamic allocation Local versus remote (processing power) Synchronous versus asynchronous (communication) Mutual awareness Authority sharing (static versus dynamic) Faillures Human errors System faults and failures

6 Automation versus Groupware Automation is only one way Collaborative systems is "another" one studied in the reserach domain called CSCW (Computer Supported Cooperative Work)

7 Group tasks McGrath J. E. Groups: Interaction and Performance. Prentice Hall, Inc., Englewood Cliffs,

8 Aspects of cooperation Ellis C. A., Gibbs S. J., Rein G., Groupware: some issues and experiences, Communications of the ACM, v.34 n.1, p.39-58, Jan

9

10 Group activities Calvary, G., Coutaz, J., Nigay, L. From single-user architectural design to PAC*: a generic software architecture model for CSCW. In Proc. of CHI '97. ACM, NY, USA,

11 The functional clover Communication Production Coordination (Salber et al., 95) 1. Services offerts 11

12 7 design considerations for Collaborative Visual Analytics (Heer & Agrawala, 2007) Division and allocation of work Common ground (experience, common knowledge) and Awareness Reference and deixis Incentive and engagement Identity, trust and reputation Group dynamics Consensus and decision making 12

13 Human-Computer Interaction issues Usability (performance, effectiveness, satisfaction) Interaction techniques Awareness versus privacy WYSIWIS Portholes Filtering Mark Stefik, Daniel G. Bobrow, Gregg Foster, Stan Lanning, Deborah G. Tatar: WYSIWIS Revised: Early Experiences with Multiuser Interfaces. ACM Trans. Inf. Syst. 5(2): (1987) Joëlle Coutaz, François Bérard, E. Carraux, James L. Crowley: Early Experience with the Mediaspace CoMedi. EHCI 1998: Jun Rekimoto: Pick-and-Drop: A Direct Manipulation Technique for Multiple Computer Environments. ACM Symposium on User Interface Software and Technology 1997: 31-39

14 Shared workspaces Integration by Metaphors (rooms): TeamRooms (Greenberg, 96) fusion of video communication and shared editing Clearboard (ishii, 92) 1. Services offerts 14

15 Shared workspaces Integration by Metaphors (rooms): TeamRooms (Greenberg, 96) fusion of video communication and shared editing Clearboard (ishii, 92) 1. Services offerts 15

16 Coordination on an ATC workstation The paper strip is a coordination tool 17

17 Coordination on an ATC workstation Conventions Paper strips positioning Colors 2. Coordination 18

18 Coordination on an ATC workstation Social norms information waving 19

19 Standard ARINC 661 Specification CDS : Control and Display System KCCU: Keyboard and Cursor Control Unit DU: Display Unit A380 Cockpit

20 Distributed Cognition? Collaborative System? Cooperation and function allocation?

21 No! Dependability! Fault Prevention/avoidance, Fault Tolerance, Fault Forecasting, fault Recovery

22 Faults and Failures Adapted from: Avizienis, A., Laprie, J.-C., Randell, B., Landwehr, C. Basic concepts and taxonomy of dependable and secure computing. In IEEE Trans. on Dependable and Secure Computing, vol.1, no.1, pp , Jan.-March 2004

23 Security Adapted from: Avizienis, A., Laprie, J.-C., Randell, B., Landwehr, C. Basic concepts and taxonomy of dependable and secure computing. In IEEE Trans. on Dependable and Secure Computing, vol.1, no.1, pp , Jan.-March 2004

24 Reliability Security

25 Security Reliability Fault Tolerance

26 Security Reliability Fault Tolerance Usability & UX

27 Security Reliability Fault Tolerance Usability & UX

28 Reducing Flying Crew New actors New tasks (empty the dishwasher) New interfaces New rules and regulations New communication media New training New failures

29 Reducing Flying Crew New actors New tasks (empty the dishwasher) New interfaces New rules and regulations New communication media New training New failures

30 Reducing Flying Crew New actors New tasks (empty the dishwasher) New interfaces New rules and regulations New communication media New training New failures

31

32

33

34 Automation Problem(s) 1. How are the autonomous entities represented in the user interface? 2. How (in general) to design interfaces that enable the control of semi-autonomous objects? 3. How to Design, Develop, Specify Evaluate user interfaces/interaction for CCS

35 Challenge for Automation in Safety-Critical Systems Analysis of the current context of the operations, Identification plans Decide for plan Input the plan Trigger supervisory system to execute plan Follow the execution of the plan (current execution and future execution) Interrupt plan Modify plan

36 Solutions Needed Operator as automation overseer Problem: how to come back to low tactical level (e.g. in case of system failure or degradation) How is the operator able to identify new plans or modifications? How is s/he able to build new plans or configurations? How to assess beforehand the impact of potential new plans or configurations? How to interact with current configuration?

37 Solutions Needed Operator as automation overseer Problem: how to come back to low tactical level (e.g. in case of system failure or degradation) How is the operator able to identify new plans or modifications? How is s/he able to build new plans or configurations? How to assess beforehand the impact of potential new plans or configurations? How to interact with current configuration?

38 Regina Bernhaupt, Guy A. Boy, Michael Feary, Philippe A. Palanque: Engineering automation in interactive critical systems. CHI Extended Abstracts 2011: 69-72

39 New User Interfaces Accessible

40 Apple strategy iphone, ipad, iwatch

41 Apple strategy iphone, ipad, iwatch

42 Alighieri, Dante. Divina Commedia The road to hell is paved with the head of people with good intentions Regina Bernhaupt, Guy A. Boy, Michael Feary, Philippe A. Palanque: Engineering automation in interactive critical systems. CHI Extended Abstracts 2011: 69-72

43 Thank You for Your Attention L enfer c est les autres Hell is other people Jean Paul Sartre Nobel Prize in Literature,

44 Examples

45 Introduction Initial Concepts Paradigm Shift to Higher Automation Levels Research directions Roles assignement Metaphors issues Design issues Implementation issues Dependability and Usability

46 Initial Concepts ATM is just all management activities devoted to reach an efficient flight trajectory for all airspace users Thus automation must be understood in this context as a tool used to improve ATM related processes through the use of technology ATM Flight Trajectories Manageme nt Automation Improvement of Processes through the use of Technology AUTOMATION IS ABOUT HUMAN VS. MACHINE (Replacement) HUMAN - MACHINE COOPERATION (Team)

47 Clock exampl e Counting minutes, second and hours difficult for human Especially when they sleep A watch is of great help for that Is this automation of a human task (no human has been doing that counting springs, using sun, ) Is this an autonomous entity

48 Clock exampl e Counting minutes, second and hours difficult for human Especially when they sleep A watch is of great help for that Is this automation of a human task (no human has been doing that counting springs, using sun, ) Is this an autonomous entity

49 The Paradigm Shift SESAR will help create a paradigm shift 1 Shift from Airspace Based operation towards a Trajectory Based operation concept. 2 Shift from Tactical Management towards a more Proactive system. 3 Shift from a controller-based system towards a more distributed system.

50 Roles Assignment w role assignment based on : First dimension BEST TIME for decision making: Strategic vs. tactical planning layer Second dimension DECISION PLACE : Controlled vs. autonomy. Third dimension BEST PLAYER : Human vs. automated player.

51 Best Decision Time (timetable) Considering that the overall system performance is taken as the main driver to decide when the decision or event should be initiated: Will the overall system maintain the required stability under the decided best time? This is important, because in an evolving ATM system during the various stages of the planning, buffers and feedback loops are necessary to react on unexpected changes. What is the impact of uncertainties in a system when most decisions are taken a long time in advance? E.g. how can this system react on a late passenger? Would it become mandatory to act like those airlines which won t wait, if you re late? How can an adaptive system be designed where the degree of strategic decisions can be chosen, e.g. depending on uncertainty and/or others factors? Do the greater number of functions allocated at strategic planning layers imply more complex and rigid operational scenarios? Can ATM system deliver required safety and efficient when most of decisions are allocated at tactical level?

52 Best Decision Place (centric vs autonomous) Considering that the overall system performance is taken as the main driver to decide where the decision or event should be initiated : What is the level of correlation between complexity and centric processes? What is the level of correlation between autonomy and centric process? To what extend are segregated airspace structures the solution to the questions of where and when to implement autonomy? Do tactical decisions imply autonomous and fully automated processes? Is high traffic density/complexity a key factor limiting autonomy? In which scenario (centric or autonomous) will automation provide higher overall system performance? Does strategic decision making imply centric scenarios?

53 Best Decision Player (human vs sytem) Considering that the overall system performance is taken as the main driver to decide where the decision or event should be initiated: Should trajectory management (e.g. Trajectory deconfliction, even tactical decisions) be fully automated? To what extent do strategic decisions require human intervention? How can uncertainty be managed in automated systems? Are the current frameworks for automation, cognition and human factors enough to capture ATM singularities? Is a fully automated air transport system socially/psychologically acceptable? Can the ATM system be decomplexified through automation? How to deal with transition issues when implementing higher levels of automation? How can resilience be taken into account in automated systems design? Does uncertainty require human centred decision making?

54 The Horse Metaphor (input)

55 A Controlable Horse (output) Look at the ears, the body shape and the tail (not easy)!!

56 A Controlable Horse (output) Look at the ears, the body shape and the tail Listen at the noise from horse Fuse that information from different sources (not easy)!!

57 A Controlable Horse (output) Look at the ears, the body shape and the tail Listen at the noise from horse Fuse that information from different sources (not easy)!!

58 The Horse Metaphor

59 The Horse Metaphor

60 The Horse Metaphor

61 The Horse Metaphor

62 The Details of the Metaphors

63 The Details of the Metaphors

64 The Details of the Metaphors

65 Number of papers on automation ( ) Num Year

66 Number of papers on automation ( ) > >

67 Dependability Issues for Automation Things will degrade Make sure they degrade gracefuly Things will fail Make sure they will fail safely Operations will carry on Make sure they carry on with good performance

68 Interactive cockpits architecture Example: the engagement of the auto-pilot through a click on the corresponding PicturePushButton Wich widget? A661_WidgetEvent Modify display A661_SetParameter_ON A FAULT-TOLERANT SOFTWARE ARCHITECTURE AND ITS FORMAL SPECIFICATION FOR EMBEDDED, REAL-TIME INTERACTIVE SYSTEMS ERTS² 2014 TOULOUSE, FRANCE 69

69 Functional Failures modes Control flow failures No Unforeseen Wrong control control control Display flow failures No Unforeseen Wrong display display display A FAULT-TOLERANT SOFTWARE ARCHITECTURE AND ITS FORMAL SPECIFICATION FOR EMBEDDED, REAL-TIME INTERACTIVE SYSTEMS ERTS² 2014 TOULOUSE, 70 FRANCE

70 The FCU FCU: Flight Control Unit Airbus A380 Flight deck A FAULT-TOLERANT SOFTWARE ARCHITECTURE AND ITS FORMAL SPECIFICATION FOR EMBEDDED, REAL-TIME INTERACTIVE SYSTEMS ERTS² 2014 TOULOUSE, FRANCE 71

71 The FCU Backup FCU: Flight Control Unit FCU Backup Airbus A380 Flight deck A FAULT-TOLERANT SOFTWARE ARCHITECTURE AND ITS FORMAL SPECIFICATION FOR EMBEDDED, REAL-TIME INTERACTIVE SYSTEMS ERTS² 2014 TOULOUSE, FRANCE 72

72 Usability issues in ATM The Human-System interaction does not exists Human-System interaction takes place using an interactive system More complex than the other systems Hardware/software integration Rapidly evolving

73 Usability issues in ATM The Human-System interaction does not exists Human-System interaction takes place using an interactive system More complex than the other systems Hardware/software integration Rapidly evolving

74 Design Issues for Automation

75 Thank You for Your Attention L enfer c est les autres Hell is other people Jean Paul Sartre Nobel Prize in Literature,

76 Remote collaboration 77

77 Domains involved HCI Distributed systems Sociology Anthropology Socail psychology Computer Science Networks and telecom Automation and autonomous entities Human Sciences Psychology Linguistics Communication Sciences Ethnomethodology B. Gaver Situated action L. Suchmann Distributed Cognition Ed. Huchins Activity Theory S. Boedker Methods Prototyping (low and high fidelity) Lab studies (experimentations) Field studies (observation) 78