This is an old revision of the document!
How to evaluate RGBD-SLAM with our datasets?
This page describes how to evaluate RGBD-SLAM using our Kinect datasets.
Code download and compilation
$ cd ~/ros $ svn checkout https://svncvpr.in.tum.de/cvpr-ros-pkg/trunk cvpr-ros-pkg $ rosmake rgbdslam_eval --rosdep-install
Dataset download and preparation
$ mkdir -p ~/data $ cd ~/data $ wget http://www9.in.tum.de/~sturmju/dirs/rgbd_datasets/freiburg1/rgbd_dataset_freiburg1_xyz.bag $ wget http://www9.in.tum.de/~sturmju/dirs/rgbd_datasets/freiburg1/rgbd_dataset_freiburg1_xyz-groundtruth.txt $ rosrun rgbd_benchmark_tools add_pointclouds_to_bagfile.py ~/data/rgbd_dataset_freiburg1_xyz.bag --nth 1
Run RGBDSLAM
$ rosrun rgbdslam_eval rgbdslam ~/data/rgbd_dataset_freiburg1_xyz-points.bag ~/data/rgbdslam_xyz.txt $ cat ~/data/rgbdslam_xyz.txt
Evaluate the trajectory versus the ground truth
$ rosrun rgbd_benchmark_tools evaluate.py ~/data/rgbd_dataset_freiburg1_xyz-groundtruth.txt ~/data/rgbdslam_xyz.txt