-
Notifications
You must be signed in to change notification settings - Fork 9
Visual SLAM Results
S(apienza) R(obust) R(obotics) G(roup) edited this page Oct 19, 2020
·
4 revisions
In this page we will collect a set of comparative experiments for our Visual-SLAM pipeline (both stereo and RGB-D). The comparison has been performed with the following state-of-the-art systems:
- ProSLAM (open-source and available at this link)
- ORB-SLAM2 (modified version to compile on GCC 7 available at this link)
We used the open-source python tool EVO to generate all the following evaluation and plots.
The dataset used are KITTI and EuROC, using only their respective stereo images. We converted such datasets from their original data format into our BOSS format and made them available at the following links:
We also developed some useful tools for conversion from our format into rosbag available here (srrg2_core_ros/src/app/srrg_to_rosbag_converter_app.cpp)