|
|
|
|
Uncertainty Management |
|
|
|
|
|
|
|
|
|
- Risk Management Consultants Ltd, UK
- Deep Blue s.r.l., Italy
- Eurocontrol Experimental Centre (EEC)
|
|
|
Introduction and Objectives |
|
Much of the information used in air traffic management (ATM) is inherently uncertain, most obviously where it involves predictions of future system states. The main approach to this uncertainty has been to reduce it at source, by developing more accurate and reliable tools and by standardizing operating procedures. However, this approach alone may not be sufficient in the future as new ATM concepts, with increased reliance on predicted information, are introduced. There may be situations in which presenting information about uncertainty to users provides additional safety and operational benefits.
A preliminary study therefore considered at a high level, whether, and if so how, uncertain information should be presented to controllers or pilots. The study included literature surveys, experience from past projects in ATM and other domains, analysis of incident data and anecdotal evidence, structured interviews and discussions with commercial pilots during flight deck observations. These tasks identified uncertainties which could potentially have a significant effect on the safety or efficiency of ATM; organisational and individual attitudes to uncertainty and ways in which uncertainty is currently presented to users.
A follow-on study was performed according to a project plan, asking where, and if so how, uncertainty could usefully be presented. To provide a practical context for these questions, three controller tools have been considered as case studies: the Tactical Load Smoother (TLS), Medium Term Conflict Detection (MTCD) and the Conflict Resolution Assistant (CORA).
|
| |
 |
A preliminary study |
| |
|
Project plan |
|
|
A process has been developed for identifying scenarios in which there could be safety or operational benefits from presenting uncertainty, and for envisioning presentation solutions. This process was applied to the case study tools and visualisations were developed and evaluated. Although this has been a relatively small-scale study with limited scope, stakeholder reaction suggests that it has raised awareness of the topic, and asked pertinent questions.
|
|
|
TLS – Radical Solutions for New Controller Roles |
For TLS, radical solutions were envisioned, involving the definition of user roles and interactions and the design of the physical workspace, as well as the graphical displays. These solutions were evaluated and refined in close co-operation with users, and it is therefore believed that they have real potential to deliver ATM benefits. However they have not been linked to real uncertainty data or integrated into realistic operational simulations.
|
|
|
MTCD/ CORA – Adding Uncertainty Information to Existing Displays |
For MTCD/ CORA, a more traditional design approach was taken, in keeping with the constraints posed by the more advanced state of development of these tools. Whereas the work on TLS sought radical solutions, the emphasis here was on more practical issues, such as obtaining appropriate data and models to drive displays, and implementing the new ideas in existing evaluation facilities.
Two examples of uncertainty connected with MTCD/ CORA were selected: the uncertainty of conflicts as seen in the MTCD Potential Problem Display, and the uncertainty in aircraft climb/ descent as seen in the Vertical Aid Window (VAW). Display solutions were sketched and implemented in the ERC Early Demonstration and Evaluation Platform (eDEP) and evaluated in the EUROCONTROL Human Factors Lab. The general opinion of the participating controllers was that the display of uncertainty, at least as presented here, provided very little benefit, although participant unfamiliarity with MTCD itself may have contributed to their reservations.
The generally negative views of controllers about the MTCD solutions contrasts with that of those involved in the TLS study, lending support to the idea that, for an unfamiliar concept such as uncertainty, the development process is especially important, as well as its products. In the TLS study, controllers had been involved in development from the outset, rather than being presented with a new tool and asked to evaluate it.
|
|
|
Where Uncertainty Information may be helpful |
Generalising from the contrast between the findings of the TLS and MTCD studies, it is expected that displays of uncertainty information, being a radical departure from current ATM practice, are likely to be of more interest as an integrated part of future controller tools than as ‘add-ons’ to existing tools.
The usefulness of presenting uncertainty information is believed to be greater for roles with a longer-term, more strategic orientation, such as Planning Controller of Multi-Sector Planner, than for the more ‘immediate’ work of the Executive Controller. Other suggested areas in which the presentation of uncertainty information might be helpful are ground control and arrival/ departure planning and management, especially in a Collaborative Decision Making environment.
|
|
|
General Guidance on presenting uncertainty |
|
Decisions about whether and how to present uncertainty need to be made for each specific scenario of use, taking into account the operational context, the user roles, tools and procedures. Nevertheless, some important lessons have been learned that could have applications beyond the case study tools, and so provide guidance on the topic of uncertainty presentation in general. The key points can be summarised as follows:
|
- The meaning of ‘presentation’ covers much more than just graphical display – it includes methods of making uncertainty evident and allowing users to interact with it.
- An iterative, scenario-based, design method is recommended, it was found very difficult to decide in the abstract what uncertainties should be presented, and involving the users at an early stage helps to overcome the problem that uncertainty is an alien concept to most ATM users. Uncertainty needs to be discussed initially in the context of real-life scenarios that users themselves identify, and related to concrete scenarios of use as development progresses.
- There is a need to be clear about what exactly is the ‘uncertainty’ being presented. There are many possible types and interpretations of uncertainty - the uncertainty shown will be specific to the particular scenario, tools, procedures and context, and users will need to have a clear understanding of what is being presented.
- Careful consideration of the time dimension is required. Providing different visualisations of current and future situations is helpful. More generally, time-line scrolling or other means should be provided to allow the user to look at different stages in the future and so understand the time evolution of uncertainty.
- There is a preference for uncertainty to be presented only on demand, although this needs to be balanced against requirements for consistency of presentation between users sharing the same picture, and the potential advantage of continuous presentation in providing a more complete picture and better situation awareness;
- There is a preference for qualitative presentations over anything ‘too mathematical’; Simple indicators, with only a small number of uncertainty levels, for example, can also limit the additional cognitive workload associated with processing information about uncertainty.
- It is helpful to give information about the causes of uncertainty, and hence whether the uncertainty is within the control of the user or must be worked around
- ‘What-if’ tools, giving users the ability to interact with uncertainty and test different solutions, were considered helpful
- There is a preference for integrated displays and tools rather than too many separate windows.
|
|
|
|
The results were presented in a dissemination forum held together with the CREA project on 15 March 2004.
|
|
|
|
| | |
