Introduction
• Research experts have been working to create software that can be put aboard aircraft to help Air Traffic Controllers manage merging streams of traffic
• The goal is to keep the aircraft properly spaced as they arrive at the airport
• Current algorithms estimate a speed to fly so that the aircraft will be properly spaced
• Speed to fly is sent to the pilot as a recommendation• Goal of this project is to create a product for UPS to use• UPS has many airplanes coming into airports at night so
this software will help Air Traffic Controller manage the large amounts of aircraft
Background
• New technology called ADS-B enables aircraft to send trajectory data directly to another aircraft
• With this advancement, aircraft can send data that is necessary to use this data to calculate a recommended speed to fly
• ADS-B communicates very frequently, so the recommended speed to fly can be recalculated often to improve accuracy
Current Algorithms
• Uses time based spacing, how many seconds behind lead aircraft
• Time based spacing naturally compresses and expands as speed changes
• Desired spacing value is the ideal spacing (normally 120 seconds)
• Spacing error is the difference between current spacing value and desired spacing value
Current Algorithm
• A more accurate method is to save trajectory data of the lead aircraft
• Look at lead aircraft’s position at t-desired spacing seconds ago
• This position is where you want to be
• Spacing error is the difference between current position and that position
Current Algorithms
• Goal of the algorithm is to fix the spacing error
• Recommended speed to fly is the estimated speed that will correct the spacing error
• Oscillations as a result of small speed or altitude changes in the target aircraft are reflected in the recommended speed to fly
Problems
• Current algorithms present too many speed commands
• Overwhelms and distracts the pilot• Can become unstable• Constantly changing speed recommendations
are obnoxious to a pilot• Surveys have shown that pilots are more likely
to ignore the recommendations if they are constantly changing or oscillating frequently
My Project
• Reduce the number of speed commands to a manageable number
• Do this while maintaining proper spacing
• Make an Improved Speed Calculation algorithm
• Create filters and quantizers to reduce the number of commands
Improved Speed Calculation
• When the lead aircraft makes a large speed change, the normal algorithm takes many small consecutive steps (blue line)
• This can be reduced to one large command (green line)
Improved Speed Calculation
• Start at t-desired spacing• Calculate the acceleration at that point• If the acceleration is above a threshold,
then the change is large enough that we will run the improved algorithm
• Progress to next data point• Run the algorithm while the acceleration at
the current data point is above the threshold
Improved Speed Calculation
• When the algorithm stops– Calculate duration: difference between time at data
point where it stops and t-desired spacing– Calculate magnitude of change: difference speed at
data point where it stops and speed at t-desired spacing
• Recommended speed to fly = current speed + magnitude of change
• Recommended speed to fly stays displayed for “duration” seconds
Experiments
• MITRE has a flight simulator in the Air Traffic Management Lab
• We recorded a flight from start to finish• Ran a simulation where that flight was used as
the lead aircraft and we were the trail aircraft• Simulation featured a focus on human factors
such as reaction time• 45 minute real-time simulations run 10-15 times• Fast time simulations run 5-10 times
Experiments
Above: The cockpit of MITRE CAASD’s flight simulator in the ATM lab
Left: Speed Recommendations are displayed on the left side of the CDTI
Results
• We consistently observed a reduction in speed recommendations of 50%
• During major speed changes, what would normally be 9-12 recommendations were reduced to 1 large recommendation
• Even when speed changes were not large enough for the improved algorithm to run, the filters and quanizers significantly reduced the number of recommendations
Implications
• During standard arrival procedures, aircraft significantly reduce speeds
• Prior to the Improved Speed Algorithm, simulations could not include arrivals because the results were extremely unpleasant
• This breakthrough allows MITRE to run a broader range of tests from departure to arrival without interruption
• This is a major step to preparing the product for field testing
Reflections
• Extremely valuable to see how computer science and engineering can be applied to a real life product
• Advice: read carefully, try to think about things in a larger scope while maintaining a detailed understanding
• Work Environment: There were several TJ grads at MITRE so it was very easy to relate to people. No real adjustment was necessary other than maintaining acting professional
• Remember, you’re representing yourself, the school, and your demographic!
Acknowledgements
• H. Peter Stassen
• Matthew Pollack
• Kelley Connolly
• Urmila Hiremath
• Mr. Latimer
• Mr. Pearce
Works Cited• EUROCONTROL, ”CoSpace 2005-ASAS Sequencing and Merging:
Flight Deck User Requirements Version 2.1”, EUROCONTROL, 2006.
• E. Hoffmann, N. Pene, K. Zeghal, ”ASAS Spacing User Requirement Document”, EEC document version 2.0
• I. Grimaud, E. Hoffman, L. Rognin, K. Zeghal, ”EACAC 2000 Real-Time experiments: Pilots perspectives”, EEC Report version 3.0
• EUROCONTROL/FAA, ”Principles of Operations for the Use of Airborne Separation Assurance Systems”, EUROCONTROL/FAA Cooperative R&D Edition 7.1, 2001
• J. Hammer, ”Preliminary analysis of an approach spacing application”, FAA/Eurocontrol R&D committee, Action plan 1, ASAS Technical Interchange Meeting, 2000