PixArt PAW3902 optical motion tracking driver #12576
Conversation
There was a problem hiding this comment.
The PR looks good to me! Where did you buy this sensor? I was not able to find any breakout.
For the pmw3901 the proportional factor was measured experimentally by measuring some known distances and rotations. @RomanBapst If we could order one we could try to repeat those measurements and then get this flying.
|
The Thoneflow driver is also stalled on measuring the scale factor. I got it to mostly match the gyro on pure rotations, but the EKF is always unhappy. |
Maybe do some flights, then replay on a hacked version of ECL to determine what scale factor gives lowest innovations? |
|
Safe to merge for continued testing and iteration. |
This is an initial driver for the new PAW3902 (and PAW3903) optical motion tracking driver from PixArt. It's an update to the popular PMW3901 with much better low light performance. https://www.pixart.com/products-comparison/16/Optical_Motion_Tracking
Not quite flight ready, there are issues with the scaling. @RomanBapst @DanielePettenuzzo @priseborough any input on how you determined the proportional factor for the PMW3901 would be appreciated. It doesn't appear to be derived from the datasheet (although it lacks quite a lot of detail).
https://github.com/PX4/Firmware/blob/ca74b3fef11a5fb7e8347e019350f24302ecd9a9/src/drivers/optical_flow/pmw3901/pmw3901.cpp#L570