| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
|
|
|
| |
This avoids any noise introduced by the kalman filter, making landing
detection much more reliable. This patch also changes the interval to
10s so that the height bounds can be increased to 4m.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
| |
Using the longer time constant could lead to false landing detection
just after apogee, which is definitely not a good idea.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Instead of using the direct output of the kalman filter and hoping
that is quiet enough to detect landing, filter that with a long
exponential decay filter and then check to make sure that doesn't
change more than 2m in 5 seconds as a trigger for landing detection.
Tested with existing telemetrum flight logs and it correctly detects
landing in all cases.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
| |
Not that telem files are currently very useful as the kalman filter
gets completly confused by the variable steps caused by missing data, but...
Signed-off-by: Keith Packard <keithp@keithp.com>
|
| |
|
|
|
|
|
| |
Use MSL instead of AGL for detecting over-range baro sensor values.
Always trust baro sensor during descent; it'll get there eventually.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
|
|
The flight code mashed together data processing, filtering and actual
flight managament into one giant pile. Split things up so that we
have:
ao_sample.c: Sensor data processing. Reads the ring, handles calibration
ao_kalman.c: Filter the data to track the accel/speed/height values
ao_flight.c: Flight state management, specific to rocketry.
The plan is to re-use ao_sample.c and ao_kalman.c for hardware not
specifically designed for rocketry, like TeleNano.
Signed-off-by: Keith Packard <keithp@keithp.com>
|
