Docker images containing a ROS 2 based installation of RTAB MAP.
1.1K
This image was used to run the evaluations of the following paper: "V. Behret, R. Kushtanova and F. Palme, "Benchmarking a ROS 2-Based Indirect SLAM on Edge Devices," 2024 4th International Conference on Robotics, Automation and Artificial Intelligence (RAAI), Singapore, Singapore, 2024, pp. 173-180, doi: 10.1109/RAAI64504.2024.10949545."
The source code to build the docker image is available at: https://github.com/hmcvlab/docker-rtab-map
Besides the docker image itself, the parameters that were tweaked for the evaluation, are needed to reproduce the results. The parameters are listed below:
# RTABMAP parameters for all nodes: rtabmap_odometry, rtabmap, stereo_odometry
---
/**:
ros__parameters:
"frame_id": "camera_gray_left"
"subscribe_stereo": true
"subscribe_odom_info": true
"wait_imu_to_init": false
"approx_sync": false
# RTAB-Map's parameters should all be string type:
"DbSqlite3/JournalMode": 0
"Grid/3D": false
"Kp/DetectorStrategy": 2
"Kp/MaxFeatures": 1000
"Mem/BinDataKept": false
"Mem/LocalizationDataSaved": true
"Mem/UseOdomGravity": true
"ORB/Gpu": true
"Odom/ScanKeyFrameThr": 0.8
"Odom/Strategy": 1
"OdomF2M/MaxSize": 5000
"Rtabmap/CreateIntermediateNodes": true
"Rtabmap/DetectionRate": 10
"Rtabmap/MemoryThr": 10000
"Rtabmap/TimeThr": 500
"Stereo/MaxDisparity": 500
"Vis/BundleAdjustment": 1
"Vis/CorNNDR": 0.9
"Vis/EstimationType": 1
"Vis/FeatureType": 2
"Vis/MaxFeatures": 1000
"Vis/MinInliers": 6
"Vis/PnPFlags": 0
If you use this docker image in your research:
@INPROCEEDINGS{10949545,
author={Behret, Valentino and Kushtanova, Regina and Palme, Frank},
booktitle={2024 4th International Conference on Robotics, Automation and Artificial Intelligence (RAAI)},
title={Benchmarking a ROS 2-Based Indirect SLAM on Edge Devices},
year={2024},
doi={10.1109/RAAI64504.2024.10949545}
}
Content type
Image
Digest
sha256:34370f875…
Size
9.9 GB
Last updated
almost 2 years ago
docker pull hmcvlab/rtab-map