You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Utilize ICP algorithm to localise captured point clouds to target prior map
AnyMal uses ICP-tools, however cannot port to Rooster due to EULA infringement issues
AICP will be used, an in house algorithm based off of the libpointmatcher library
Features a non-slam for pure localisation which we will use
Going Forward: Short Term
Become familiar with AICP and running it on existing datasets
Make sure master branch on machine is up to date with latest code
Download data sets from drs_testing_data
Become familiar with odometry specifically visual odometry with FOVIS
Edit AICP such that we can save individual point clouds for piping into ESM, as well as saving metadata (pointcloud no, timestamps, poses, quaternions) to a csv file.
Familiarise more with Visual SLAM, Deep Learning Loop Closure and ESM
Localisation + Tools
Going Forward: Short Term