Skip to content

README.md

Latest commit

 

History

History
54 lines (32 loc) · 2.11 KB

README.md

File metadata and controls

54 lines (32 loc) · 2.11 KB

📏 Benchmarking vS-Graphs

📊 Evaluation

To evaluate vS-Graphs against other Visual SLAM systems (e.g., ORB-SLAM3), follow the steps below:

🔧 1. Run Pose Storage Script

Run generate_kf_pose_txt_files.py to store estimated robot poses (based on the KeyFrames). Make sure that you set proper configuration parameters beforehand in the config.yaml file. The script will create a .txt file and populate it with pose data:

python [workspace]/src/visual_sgraphs/evaluation/generate_kf_pose_txt_files.py

▶️ 2. Run vS-Graphs

Launch vS-Graphs and play the ros2 bag file. Pose data will be recorded automatically into the generated .txt file during execution.

📈 3. Run Evaluation

Once both ground-truth (such as LiDAR S-Graphs) and estimated poses (the generated .txt file) are ready, use evo to compute Absolute Pose Error (APE):

evo_ape tum [gt_pose].txt [vs_graphs_pose].txt -va > results.txt --plot --plot_mode xy

🗺️ Working with Maps

vS-Graphs uses .osa files to store maps. These files are saved by default in your ROS_HOME directory (~/.ros/).

🔄 Loading a Map

  • To load a saved map, set the System.LoadAtlasFromFile parameter in your SLAM settings YAML file.
  • ⚠️ If no .osa file is available, comment out the System.LoadAtlasFromFile parameter to avoid runtime errors.

💾 Saving a Map

  • Option 1: Enable System.SaveAtlasToFile in the SLAM settings file. The map will be automatically saved when you shut down the ROS node.
  • Option 2: Manually trigger a save using the following ROS service:
rosservice call /vs_graphs/save_map [file_name]

🛠️ Map and Trajectory ROS Services

Use the following ROS services to save maps and trajectories during or after running vS-Graphs:

  • To save the current SLAM map to ~/.ros/[file_name].osa use rosservice call /vs_graphs/save_map [file_name].
  • To exports the estimated trajectories use rosservice call /vs_graphs/save_traj [file_name]