Hi, I tested some mavlink precision landing code in SITL which seemed OK so I went and tried a couple of flights. I did some flights and landing first normally without the OBC connected and all was well, so then I connected the OBC. Whenever the OBC (mavproxy+dronekit script sending landing_target and distance_sensor messages) was connected it would experience large random twitches followed by inverted nosedive into the hedge.
I’ve attached one log as an example. In this log I do a couple of test flights and landings first without OBC, then connect the OBC. At 19115 I ask it to land which it starts and then shortly goes bananas - twitches then turns over and dives into the hedge. By 20310 it’s all gone pear shaped. I notice in the logs there are quite a few gaps after PL messages eg. 20425 and then there’s an EKF anomaly and unexpected event 62, whatever that is. Note that CTUN.alt carries down into negative territory even though it was sitting in a hedge at that point. It looks like there are lots of EKF innovations in the latter part of the log, but I don’t know if these are really abnormal or not.
I don’t believe the dronekit script was actually sending any messages at all during this period, but I’m adding more debug output and will try to replicate this again tomorrow.
433.BIN (723.7 KB)
Hmm, please (continue to) ignore this for now, I think this was either caused by sending dud mavlink messages or by massive EMI that is interfering with various hardware components.