Any suggestions for tuning a larger scale (outdoor) environment

[notalltim]

I am working with a data set collected from airsim (neighborhood environment shown below) Do you have any thoughts on optimizing for larger environments?

[ToniRV]

Hi @Timbama, looks nice! Kimera-VIO itself is running a fixed-lag smoothing backend, so there should be no scalability issue for pose estimation. Pose-graph optimization might be a bit more tricky, but since it is running in another slower thread, it should not be a problem. For metric-semantic reconstruction with Kimera-Semantics, it scales fairly well, but there is no current mechanism developed to save the 3D reconstruction to disk to make it truly scalable to large-scale scenarios, it shouldn't be too difficult though).