The future of Air Traffic Flow Management

(I had added this as a comment to the previous blog entry. One of the readers pointed out that this entry should be "elevated" to a real blog entry, because it was an interesting entry. So, I did.)

I wanted to relate the blog-story below of flight delays to one of our Brahms research projects at NASA. Although being delayed for a day is not fun, I do get a kick out of my research touching real-life situations. Not surprising, because in this case our research is about developing new, more collaborative ways to deal with future over-crowding of the US airspace.

In our Collaborative Traffic-Flow Decision Management project, we are studying how the FAA and the airlines can be more collaborative in deciding when and which flights to delay or cancel. Currently, the FAA decides, more or less, on their own which flights to delay. These decisions are made on a two-hour cycle in the US. The unfortunate consequence of delaying or canceling flights is that these have a rippling effect in consecutive flights, both for the airlines and for airline passengers. Our situation today is a consequence of such a decision.

How do airline crews get to their flights on time? Indeed, very often they need to fly to the destination from which the flight is to take off. When a flight is delayed, or cancelled, this means that the crew onboard these flights will also be delayed. Thus. this ripple-effect does not only delay the flight the crew is on, but also the flight that they have to be on next, etc. Often the airlines bring in a "reserve" crew, but this does not always work.

What we are researching, using our Brahms simulation language, are ways in which the airlines can participate with the FAA in deciding what flights to delay. Currently, the airlines cannot provide the FAA with a "wish-list" of actions to take, given certain flight delays. We are using a multi BDI-agent simulation approach to simulate how both airlines and FAA will collaboratively decide which flights to delay. BDI stands for Belief-Desire-Intention and is an AI-based approach, enabling software agents to reason and act based on beliefs they have. Allowing agents to communicate beliefs enables us to use this approach to model multi-party coordination, simulating some sort of "back-and-forth" debate between airlines and FAA.

Given my experience today I wonder, as I had wondered a while back in our research, how we could get passengers to participate in this debate. Wouldn't it have been nice if I could have told the airlines how important it was for me to make my flight to Amsterdam? Maybe, I would have been willing to pay extra to make my flight. If the airlines would have known, and could have taken this priority into account, they could have convinced the FAA to make another decision.

Of course, it is not as simple as that. But, today I am thinking ... it would have been nice! Although, on second thought, I am Dutch and there would have been no way I would have paid more to be there on time ... what's in a day :-)