Initial 3dof with DepthAI
Merge request reports
Activity
- Resolved by Moshi Turner
- Resolved by Moshi Turner
It looks good to me, there are a couple of dangling comments and mock variable/field names I think, but you'll probably remove those later.
The overall separation of the data source (frame server) and the device looks good.
It's good that you got 3dof working. I would be super careful with the timestamp stuff though, those could easily break a future SLAM integration.
In particular:
- Are you sure depthai gives you timestamps in "system clock"? i.e., a hardware timestamp that is corrected taking into consideration a good (and dynamic) estimate of the USB RTT? It looks like so because of the
dai_ts_to_monado_ts
function but it's worth to double check. - Do you know if depthai frame and IMU timestamps are properly synced? If you print the timestamps of both frames and IMU samples as soon as you receive them you should see frames that are about 5ms-10ms in the past, more than that could point to an issue in the sync mechanism.
- Are you sure depthai gives you timestamps in "system clock"? i.e., a hardware timestamp that is corrected taking into consideration a good (and dynamic) estimate of the USB RTT? It looks like so because of the
added 5 commits
Toggle commit listadded 5 commits
Toggle commit list