Closed hidmic closed 2 years ago
@glpuga can you reproduce this issue? I went back to https://github.com/ekumenlabs/safe-control-gym/commit/fb0eed7a9d1290cd85e8c28fa152b75c8141959e and still.
This creates more problems than it solves. Longer take-off durations cause collisions with the floor (as the drone starts out a few cm above ground).
Just saw this. I noticed a bunch of violations too while testing. We are now running on the very latest competition code (I realized whey updating the docker to the new repo that the docker probably had been running code from a couple weeks back instead).
There may have been some update to the competition code.
I guess the answer to https://github.com/utiasDSL/safe-control-gym/discussions/106 is good for this too, I don't think we can do much about this since we need to go through take-off one way or another and it's very little we can do to improve it.
For reasons I don't fully understand, I started getting constraint violations on take-off (!). This happens on the
ekumenlabs/submission
branch for scenario 1301.Increasing take-off time reduces the overshoot. I tried to reduce the braking time to compensate, but from what I'm seeing it doesn't help much :(
Glad #5 is up.