Wizard of Oz prototyping for automated decision-making tools in air traffic control (WIZARD)
How can we assure the safety of automated decision-making tools in the complex air traffic control environment?
The challenge
Air traffic control (ATC) is a complex socio-technical environment. Introducing automated decision-making could improve the efficiency and effectiveness of the sector. But existing methods to assure the safety of such tools require simulation in dedicated physical environments with skilled personnel, specialised facilities, and equipment that are resource and time intensive to set up and deliver.
This project is asking the question, can Wizard of Oz prototyping, validation and testing, where an individual simulates the automated decision-making capabilities of the tool by controlling how the interface responds to user input, be used to develop and test a new automated decision-making prototype tool?
The research
The project will build on the outcome of the recently completed A2URE project, by testing the initial stages of a methodology for designing, developing and testing automated decision-making tools. It will develop and evaluate an innovative approach to allow for future ATC automated decision-making capabilities to be prototyped and validated in a more affordable and less time and resource-intensive way.
The prototype will then be tested with a group of air traffic controllers (ATCOs) within the Institute for Safe Autonomy’s Human-Robot Interaction lab at the University of York.
The progress
Interviews with a group of ATCOs have helped to understand their current working practices. The findings have been translated into a number of user requirements that are now informing the development of an interactive prototype of an automated decision-making tool for air traffic control operations.
To support this development, the team held workshops to explore the types of scenarios a user would encounter with such a tool, and how they might respond. This helped them gain an understanding of what the tool should look like and what it could be capable of doing.
In the next phase of the project, the team will gather feedback on how ATCOs may respond and interact with such a tool. This will be done by allowing users to interact with paper and digital prototypes to offer suggestions for improvement, as well as getting an understanding of this type of approach for validation using Wizard of Oz prototyping. In turn, this will highlight what potential hazards and safety concerns can be gained early in the development process.
- 1.2.1 Considering human/machine interactions
- Sujan, M., Pool, R., and Salmon, P. "Eight human factors and ergonomics principles for healthcare artificial intelligence" in BMJ Health & Care Informatics (January 2022)
- A panel discussion on “The role of HF/E in AI operation and evaluation” was held at the Ergonomics & Human Factors Annual Conference in April 2021
Project partners
- Dr Jo Iacovides (principal investigator) University of York
- Preetam Heeramun (principal investigator) NATS