If this proves feasible, there could be a consulting/subcontracting opportunity for a Flightgear developer/SME to assist us scenario development for our project. Hopefully that makes sense! I greatly appreciate anyone taking the time to consider this and offer any insight on the feasibility of using Flightgear to generate the simulated data for the above scenario, and the associated approach and evel-of-effort for this concept. Evaluating these errors for different sensors against different targets for different types of aircraft would be the basis for developing a scorecard building block for each type of aircraft. The simulated sensor data would allow us to incorporate position error (lat, long, altitude change) and velocity error (course, speed, vertical change) that could then be evaluated against the known ground truth location of each target (based on the actual simulation scenario). Sensor data from F-16’s tracking enemy aircraft and ground based radar locations (SAM sites) Sensor data from AWACS tracking enemy aircraft and ground based radar locations (SAM sites) Flight tracks of two “Friendly” Aircraft With the above context in mind, I am evaluating IF and HOW Flightgear could be used to generate flight track and sensor data that would represent the attached scenario (see screenshot on this link)ĪIRCRAFT FLIGHT DATA (Timing, position, velocity) By developing and validating basic scorecard building blocks with relevant, clean simulated data, we could accelerate development and the increase confidence in results when running the code on actual complex, classified data sets. Info on the ORANGE FLAG exercises that AETC conducts quarterly:ĬREATING SIMULATED DATA WITH FLIGHTGEAR: We are evaluating the ability to develop synthetic/simulated data using Flightgear (or other open source tool) that could represent basic operational scenarios and enable rapid development and testing of new code. Ground based radar sites, enemy aircraft, missiles in flight, etc)ĬHALLENGE: The actual data collected from aircraft during test exercises is often incomplete and inconsistent in format, contains a huge amount of extraneous data, and is classified at many different security levels (restricting access for code development). We are evaluating the ability to apply machine learning to these huge data repositories to develop “scorecards” that evaluate the accuracy of different types of sensors on various aircraft to identify and track targets (ie. AETC has huge amounts of data that it collects from every sensor system on every type of aircraft in the US inventory (and foreign aircraft as well). I have outlined the context and concept below.ĬONTEXT: We are working with the Air Force Engineering Center to develop analytic techniques to evaluate the accuracy (“truthfulness”) of various sensor systems to identify and track targets in complex operational environments. I am new to Flightgear and this forum, and am looking for assistance re a project I am working on for the US Air Force.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |