I have a Raspberry Pi connected to my mount and cameras running indiserver - I connect to it with KStars and EKOS from my laptop (over wifi - I know this isn't ideal). Most often I get pinpoint stars and it works great, but some times I also get movement even though guiding is running as expected.
My jobs are setup to break if guiding has above 5" deviation - so I'm puzzled to get pictures like the ones attached.
What type of scope are you using SCT, refractor, etc?
Any idea of approx AZ of M81 when image was taken?
Do you have any flex between guide and main scope?
I took your image and did a plate solve on it, shows about 30arcsec of 'tadpoles', see attached.
This looks like a 'jump' happened.
When image starts, has slew and guider been stable for a number of seconds prior to image taking?