On Tuesday 14 April 2009, Curtis Olson wrote: > I'm involved in a project where we are developing, refining, and > tuning a fairly advanced algorithm that fuses inertial data (gyro > & accelerometer) with gps data to produce a pretty accurate > attitude and location estimate. (The algorithm will be released > as open-source.) One difficult part of this project is that when > we are working with real world sensor data and real world > aircraft (or UAV's) we have no "truth" reference. Our algorithm > produces an answer, but we have no idea exactly how close we are > to the truth because we don't have any way to record the true > location and attitude on a real flight. (And even if we did have > a way to record the flight parameters, we would be trying to > match more accurate sensors, and better algorithms, but still not > the "truth".) > > FlightGear is useful for truth testing because we can fly some > flight profile and record the sequence of all the exact locations > and attitude angles. We can also record all the "perfect" sensor > readings and later optionally add noise and other real world > artifacts. This way we can create a realistic data set of sensor > values along with the true location and attitude at every time > step for an entire flight profile. We can run the data through > our algorithm and work on tuning and refining it to produce > optimal results. > > This is all very nice, but it would be cool to use FlightGear as > a visualization tool for the final result. What would be really > nice is to animate/replay the "true" flight of the aircraft and > then superimpose the estimated aircraft position and orientation. > One way to do this would be to create a version of an aircraft > that has a second copy of itself with the ability to offset it's > position and orientation from the true model. > > My questions is this ... from a modeling perspective, can that > 2nd aircraft be animated with absolute lon/lat/elev and > roll/pitch/yaw degrees? Or would we need to compute an X, Y, Z > offset in meters for the second aircraft? It would be a pain to > figure out the orientation transform relative to the original > aircraft ... can the secondary aircraft be animated with absolute > angles relative to the world coordinate frame? > > (For animating the 2nd aircraft, I imagine we'd create some set > of custom property names, and I'd drive them externally via some > network/file protocol ... maybe creating a custom configuration > for the "generic" protocol.) > > Thanks, > > Curt.
I think that if you're trying to do it via animation you can only do it relative to the main aircraft position and attitude because all the aircraft animations are relative to its frame of reference; there's no absolute frame of reference where you could specify an absolute location or attitude. This is the method I used for the old and now defunct SeaHawk-pair, as well as for high-elevation terrain markers and navigation target markers. Working out the relative animations isn't too difficult though; if you've got an absolute lat/lon it's easy enough to find the relative offsets using the Nasal Geo functions. LeeE ------------------------------------------------------------------------------ This SF.net email is sponsored by: High Quality Requirements in a Collaborative Environment. Download a free trial of Rational Requirements Composer Now! http://p.sf.net/sfu/www-ibm-com _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
