shavera

Feature request: gradient descent for objective While usually perform this task in Maneuver Node Controller, as a functionality, it feels more appropriate for Flight Plan Use Case: I have a maneuver node with a vector approximately intercepting a target. (Perhaps for a variety of reasons the orbital transfer/rendezvous calcuation isn't going to work.) Right now, I'll pull up the Maneuver Node Controller window and pin the 'distance to intercept' window, and then manually adjust the vector, increasing or decreasing each component until the distance begins to increase, then cycling through the other components. It would be nice to be able to have a gradient descent tool that does that for me, tweaking the vector until the closest approach distance (may need to specify the c1 or c2 approach) is minimized. As an aside, this would pair nicely with my scale/magnitude suggestion in the Maneuver Node Controller thread. Rather than tweaking each vector component with possibly very different step sizes, tweaking the overall direction and magnitude of the vector may be a better basis space for the descent.

Feature request: Scale and Rotation Mode Use case: using an online transfer calculator may produce a delta-v vector that assumes instantaneous impulse. However, perhaps the engine burn time is so long that the assumption no longer holds (cosine losses, e.g.) It might be helpful to input the vector as (prograde, normal, radial) components, but then be able to scale the magnitude of the vector while keeping the same direction. Use case: I have more dv than I need for a transfer, and I know that overall I can afford <some overall magnitude>. I would like to be able to set that magnitude, then play around with rotating the vector until it creates the trajectory I want. Use case: I have a maneuver node with approximately the right vector, but I'd like to fine tune it. However, given the scale, perhaps the prograde component is significantly dominant over the other two components. In order to fine-tune the vector I need to adjust the other components by larger increments (say, on the order of 10s of m/s), but the prograde component by small increments (1 m/s); as the vector rotates, the relative adjustment scales to each component change. Instead, if I have a vector more or less in the right direction, and I can simply use rotations and scaling to fine tune the vector more smoothly. If I knew C# I'd try to program it and PR it, but it will be a minute til I can figure that out.