Odom publishes apriori estimate?

[pratro]

Dear Koide3,

Is there a reason why the callback to on_new_frame happens before the fused estimate is computed? (before call to update_smoother)

https://github.com/koide3/glim/blob/46c489e345a38ae6c06fb136000d0ce10d03fc7b/src/glim/odometry/odometry_estimation_imu.cpp#L307

The transforms in the above frame passed to Callbacks:: contain only a forward propagation (till the current frame's timestamp) using IMU since the last frame's timestamp but the current frame's points (ICP factor), IMU factor and Bias aren't yet fused to provide a transform estimation. So only an apriori is published via this callback (that leads to odom publishing), is that interpretation correct?

Thank you

[koide3]

Your understanding is correct. The current implementation publishes the result of IMU-forward propagation as /glim_ros/odom. I expected it would be good for scenarios that require low-latency outputs (e.g., path planning) with reasonably small estimation drift.

Meanwhile, I'm thinking of implementing one more output topic to publish poses corrected by point cloud matching for people who need more precise pose outputs.

[pratro]

Thanks for your speedy response. I see. Yes a precise pose (posteriori) publisher is a great idea. Do you anticipate this could come without the latency penalty?