Our implementation of voxel and BEV mapping, given lidar and camera.
ROS2: ros2 launch physics_atv_visual_mapping voxel_localmapping.launch.py
Python example can be found in scripts/offline_processing/postprocess_tartandrive.py
Example config can be found in config/ros/super_odometry_integration.yaml
| Topic | Msg Type | Description |
|---|---|---|
/dino_image |
sensor_msgs/Image |
A visualization of the image features extracted from the raw image |
/dino_pcl |
sensor_msgs/PointCloud2 |
The pointcloud, colorized with the VFM feature visualization |
/dino_voxels |
sensor_msgs/PointCloud2 |
The voxel grid with all visual features |
/dino_voxels_viz |
sensor_msgs/PointCloud2 |
The voxel grid with visualization features |
/dino_gridmap |
grid_map_msgs/GridMap |
A BEV-projection of the voxel grid. Only publishes if terrain estimation is enabled |
| Topic | Msg Type | Description |
|---|---|---|
image_topic |
sensor_msgs/{Image/CompressedImage} |
The topic to get images from (configurable) |
camera_info_topic |
sensor_msgs/CameraInfo |
The topic to get intrinsics from (configurable) |
pointcloud_topic |
sensor_msgs/PointCloud2 |
The topic to get pointclouds from (configurable) |
odometry_topic |
nav_msgs/Odometry |
The topic to get odometry from (configurable). The map will be in this message's base frame |