Changelog for package ainstein_radar_tools

3.0.1 (2020-02-11)

3.0.0 (2020-02-06)

• Major refactor, add conversion header and nodelets Refactored the conversion utilities to live within a namespace instead of the radar to pointcloud class, changed their usage in all dependent files. Added nodelets for the passthrough and radar to pointcloud filters, tested on K79 data. Removed old nodelets which weren't being built properly.
• Fix radar camera fusion output image and launches Fixed the name of the radar+camera fusion class' output image topic to be scoped within a new private image transport instance, and fixed the launch files to use the correct topic name.
• Minor, fix launch for radar camera val
• Minor fixes to radar+camera fusion launch and node Fixed the radar+camera fusion launch file to use the updated topic names for radar and camera data. Also fixed the fusion class itself to prevent crashing when empty bounding box arrays are processed. This node is still intended for use with the tracking filter.
• Fix multiple target rendering without SNR alpha Fixed rendering multiple target rectangles when the SNR-based alpha is not used for blending. Now renders all targets instead of only the first one. Also fixed a few small issues with other files.
• Contributors: Nick Rotella

2.0.2 (2019-11-19)

• Use RadarInfo for sizing validation 2d bounding box Changed the radar-camera validation node, which draws 2d bounding boxes on the input image corresponding to the radar sensor's specifications, to use the actual RadarInfo message assumed to be published by any radar which publishes data. Needs testing on hardware.
• Separate radar camera validation class from node
• Rename radar camera test node, update launch files Renamed the radar camera "test" node to rdara camera "validation" and updated launch files for T79 and added one for K79. Testing again with K79 to verify this still works and get screenshots for wiki tutorials. In the future, should separate radar camera validation class from the node for portability, same as radar camera fusion class/node setup.
• Contributors: Nick Rotella

2.0.1 (2019-11-12)

• Add vision_msgs as ainstein_radar_tools dependency
• Contributors: Nick Rotella

2.0.0 (2019-11-12)

• Add changelog for new subpkg ainstein_radar_tools
• Add 3d bounding box output from radar camera fusion Added 3d bounding box publishing from the radar camera fusion class which uses the radar tracking filter 2d bounding box (assuming it is published) to get the width and depth of the object and uses the object height from the object detector (optionally also uses the object detector reported width instead of radar data). This is done by projecting the 2d image bounding box into 3d space at the distance of the tracked target.
• Add working radar/camera fusion using TensorFlow Added a working radar/camera fusion or "cross-validation" class which annotates objects detected from a camera image using a pre-trained TensorFlow-based 2d object detector with radar information for all detected objects which overlap with radar data. Functionality only has "runtime dependencies" on the TensorFlow object detector in the sense that fusion is driven by radar, camera, and detected object callbacks. The fusion node is also prevented from running until the object detector node advertises a service indicating that it's ready. Finally, the label map from object index to string name is expected to be set in the parameter server as a dictionary by the object detector. The object detector itself could therefore be anything which outputs vision_msgs/Detection2DArray messages, advertises an "is ready" service and sets the label "database" (map) in the parameter server. For an example on how to use T79 with a RealSense d435 (RGB camera only) and set the correct topic/service/parameter mappings, see the launch file added in this commit.
• Developing radar+camera cross-validation "fusion" Testing a new node for radar+camera cross-validation using pre-trained TensorFlow models for 2d object detection combined with radar data to display bounding boxes associated with radar detections. WIP.
• Add new ainstein_radar_tools subpkg Added a new ainstein_radar_tools subpackage to ainstein_radar which is meant to store tools and utilities based on the other subpackages but not core to development, for example sensor fusion and SLAM nodes using radar data among other sensors. This could arguably be broken out into its own package and will be if necessary, however the intent is for these tools to aid in development for anyone using Ainstein radars. The first and only tool in this subpackage is a simple replacement for the "CapApp" radar/camera sensor fusion application which draws boxes over the image to indicate targets. This requires a calibrated camera publishing CameraInfo messages (a RealSense d435i was used for the development). A sample launch file for using this with T79 is in the launch folder and will probably be replaced with a ROS wiki tutorial.
• Contributors: Nick Rotella
• Add 3d bounding box output from radar camera fusion Added 3d bounding box publishing from the radar camera fusion class which uses the radar tracking filter 2d bounding box (assuming it is published) to get the width and depth of the object and uses the object height from the object detector (optionally also uses the object detector reported width instead of radar data). This is done by projecting the 2d image bounding box into 3d space at the distance of the tracked target.
• Add working radar/camera fusion using TensorFlow Added a working radar/camera fusion or "cross-validation" class which annotates objects detected from a camera image using a pre-trained TensorFlow-based 2d object detector with radar information for all detected objects which overlap with radar data. Functionality only has "runtime dependencies" on the TensorFlow object detector in the sense that fusion is driven by radar, camera, and detected object callbacks. The fusion node is also prevented from running until the object detector node advertises a service indicating that it's ready. Finally, the label map from object index to string name is expected to be set in the parameter server as a dictionary by the object detector. The object detector itself could therefore be anything which outputs vision_msgs/Detection2DArray messages, advertises an "is ready" service and sets the label "database" (map) in the parameter server. For an example on how to use T79 with a RealSense d435 (RGB camera only) and set the correct topic/service/parameter mappings, see the launch file added in this commit.
• Developing radar+camera cross-validation "fusion" Testing a new node for radar+camera cross-validation using pre-trained TensorFlow models for 2d object detection combined with radar data to display bounding boxes associated with radar detections. WIP.
• Add new ainstein_radar_tools subpkg Added a new ainstein_radar_tools subpackage to ainstein_radar which is meant to store tools and utilities based on the other subpackages but not core to development, for example sensor fusion and SLAM nodes using radar data among other sensors. This could arguably be broken out into its own package and will be if necessary, however the intent is for these tools to aid in development for anyone using Ainstein radars. The first and only tool in this subpackage is a simple replacement for the "CapApp" radar/camera sensor fusion application which draws boxes over the image to indicate targets. This requires a calibrated camera publishing CameraInfo messages (a RealSense d435i was used for the development). A sample launch file for using this with T79 is in the launch folder and will probably be replaced with a ROS wiki tutorial.
• Contributors: Nick Rotella

1.1.0 (2019-10-29)

1.0.3 (2019-10-03)

1.0.2 (2019-09-25)

1.0.1 (2019-09-24)