Hello,
We have been testing a brand new drone with PX4 v1.14.0 and a Holybro 6X autopilot. The drone has a motor to motor distance of 690mm and a total weight of about 3.8kg
We first did a basic and fast PID tunning (yaw was not tunned) and tested the performance in Stabilized, Altitude and Position modes. Drone seemed stable all the time and had a nice overall performance.
After that, we decided to test the PID Autotune feature and compare the given values.
To us, Autotune seemed to provide _P values that were too high, creating fast oscillations in Stabilized mode when some fast movements were commanded.
While inspecting the logs, it seemed like pitchrate and rollrate was overshooting by a little bit.
The differences between our basic tunning and the Autotune are the following:
| MC_PITCHRATE_D | MC_PITCHRATE_I | MC_PITCHRATE_P | MC_PITCH_P | MC_ROLLRATE_D | MC_ROLLRATE_I | MC_ROLLRATE_P | MC_ROLL_P | MC_YAWRATE_D | MC_YAWRATE_I | MC_YAWRATE_P | MC_YAW_P |
|----------------|----------------|----------------|------------|---------------|---------------|---------------|-----------|--------------|--------------|--------------|----------|
| 0.0017 | 0.1199 | 0.0900 | 7.0 | 0.0013 | 0.0900 | 0.0850 | 6.7500 | 0.0000 | 0.1000 | 0.2000 | 2.7999 |
| 0.0031 | 0.2000 | 0.1459 | 6.5000 | 0.0007 | 0.1749 | 0.1278 | 6.5000 | 0.0009 | 0.1599 | 0.1269 | 6.5000 |
First row is our tunning, second row has the Autotune values. K is1 in both cases.
Right now I can only share two logs:
- Our custom tunning: https://logs.px4.io/plot_app?log=3ca02426-1274-4485-ace9-3793f176db78
- Autotune tunning: https://logs.px4.io/plot_app?log=6d82067e-9dc7-4663-9b82-930176910b42
I would like to get a report or evaluation on how good or bad you think the Autotune has performed and hopefully it can provide you with more analysis data. Thanks.