External vision yaw being rejected for portions of the flight

[marcelino-pensa]

Hi,

I am flying px4 1.12.3 with mavros 1.13.0 providing external position and heading from a T265 module.

In one of my flights I had noticed that px4's heading drifted apart from the external vision yaw (blue plot has external vision heading, red plot has px4 heading):

I went ahead to inspect the ulog, and I have noticed that the flag reject_yaw did turn to true midway through this flight:

Then I checked the heading innovation, and it does shoot up quickly, then it slowly moved back down, never reaching a point where it is low enough to allow external vision yaw to be used:

Now I'm wondering if there is any way to disable this innovation test. I want the estimator to always trust external pose. What I imagine is that I need to increase some standard deviation parameters. I looked into the EKF code, and the variable for reject_yaw only shows up in gps_yaw_fusion and mag_fusion, but that is confusing me, as I'm not using either of those.

My question is: can I disable such test? Should I just change the standard deviation for one of these parameters? Which one? I could only find EKF2_HDG_GATE (which is supposed to be used with a magnetometer) and EKF2_MAG_GATE (also for magnetometer).

Note: my current value for EKF2_AID_MASK is 280.

[marcelino-pensa]

Oops, forgot to upload the ulog. Here it is: log_257_2022-3-11-13-35-38.zip

[Xiaoye-A]

您好，奎已经收到你的邮件，谢谢您的来信哦！奎会尽快给您回复。