Changelog for package ainstein_radar_drivers

3.0.1 (2020-02-11)

3.0.0 (2020-02-06)

• Merge branch \'master\' of https://github.com/AinsteinAI/ainstein_radar
• Minor, needed to negate speed for K79-3D using new SNR firmware
• Add SNR to K79 and K79-3D output Added SNR output corresponding to the new fimrware which packs the SNR in the message in the expected place. SNR sppears to be from roughly 140-200, with some targets at 0 (needs to be debugged).
• Minor, update K79 specs to longer max range
• Minor, add missing K79 specifications
• Contributors: Nick Rotella

2.0.2 (2019-11-19)

• Minor, remove temporary file
• Minor, fix SRD-D1 output data topic name
• Contributors: Nick Rotella

2.0.1 (2019-11-12)

2.0.0 (2019-11-12)

• Update RadarInfo msg, refactor for K79 and add T79 Updated the RadarInfo message slightly. Also refactored the K79 interface class to publish the RadarInfo message. Added publishing the RadarInfo message for T79 as well, needs testing.
• Add latched publishing of RadarInfo msg for K79 Added a latched publication of RadarInfo message for K79 in the K79 interface class. This message is published once and stored for all future subscribers since the radar specifications never change (as of now). The specifications used come from the preliminary datasheet and are likely to change. Any unknown specifications are left at zero.
• Contributors: Nick Rotella

1.1.0 (2019-10-29)

• Minor fixes to package XML formatting Fixed the package XML file formatting and added missing content to conform to the suggested style guidelines.
• Add simple T79 (ZOI) configuration GUI Added a simple Tk-based configuration GUI for the non-BSD T79 firmware (ZOI or Zone Of Interest firmware). This GUI allows setting the \"Nonvolatile Radar Parameters Command\" which includes CAN ID, baudrate, and self start behavior.
• Add K79-3D nodelet Added a nodelet for K79-3D to enable zero-copy message passing within the same nodelet manager. Also added a launch file for testing this nodelet. Tested on hardware and working.
• Add non-BSD T79 node for ZOI firmware/deprecate BSD Added a new interface and node for the new ZOI (Zone of Interest) T79 firmware and deprecated the old T79 BSD interface and node by moving them as needed. Both tested and working, however nodelets are not yet available due to runtime issues involving the virtual base class. The new T79 node allows setting the zone of interest (min/max range, min/max azimuth angle) via ROS dynamic reconfigure GUI. The new node also supports up to 10 radars at once, specifying the CAN ID of each as a fixed node parameter. To set the CAN ID for a radar, currently this must be done outside ROS. A service call may implement this soon.
• Bug fix in K79-3D interface, working in parking lot Fixed a few bugs in the parsing of K79-3D data packets which caused the data to be interpreted incorrectly resulting in nonphysical range and angles. Tested in the parking lot and seems to be working properly.
• Contributors: Nick Rotella

1.0.3 (2019-10-03)

1.0.2 (2019-09-25)

• Remove unused includes causing dependency issues
• Contributors: Nick Rotella

1.0.1 (2019-09-24)

• Refactor filters into separate subpkg, fix bug Created subpackage ainstein_radar_filters for radar filters and conversions, moved all filters from ainstein_radar_drivers into this subpkg and tested build and launch on rosbag data. Also fixed a small bug in the radar data range filter in which the dynamic reconfigure callback was not being registered, preventing the filter from working. Now, the filter compiles and works properly.
• Fix CMake issue and move nodelets to plugins folder Added missing dependency to range filter nodelet which was setting up a build race condition in which the dynamic reconfigure header sometimes was built after the nodelet, causing the build to fail. Also created a plugins folder and moved all nodelet XML files there, and updated the package.xml to match.
• Build fix, add radardata range filter nodelet xml Added the radardata_range_filter nodelet XML file to the package XML which fixes a build issue involving missing config for this nodelet.
• Refactor K79 interface to break out pure C++ driver Refactored the K79 radar interface to break out a pure C/C++ driver class called RadarDriverK79 on which RadarInterfaceK79 now depends. Tested and working with K79-2D, would need to do the same for K79-3D as well as all other radars which have ROS-dependent drivers. The next step would be to move the pure C++ driver outside this pkg entirely and create an external dependency, allowing use of radars without needing ROS installed.
• Merge branch \'master\' of https://github.com/AinsteinAI/ainstein_radar
• Add range filter dynamic reconfigure, needs testing
• Rename all instances of usharp3d to srd_d1 Renamed all instances of usharp3d to srd_d1 to follow new product naming. srd_d1 python node should be tested and cleaned up and drop support for old firmware going forward.
• Minor, change topic naming for radar outputs
• Minor, fix timestamps on radar data
• Minor, remove sign flip from filter, was a hack
• Merge branch \'nick-addK79-3D\'
• Add K79-3D node and fix azimuth K79-2D parser issue Added the K79 3D and verified the data, though additional testing may need to be done. Added the interface (which is very similar to K79-2D and should be combined later), node and launch file. Fixed an issue with azimuth parsing which caused targets to appear behind the sensor. This was due to the azimuth angle being read as a 16bit signed int when in fact it is only a single byte (unsigned). This change was also implemented to K79-3D.
• Small fix to K79 speed so that away is +ve
• Add K79-3D interface class and test node Added a new K79-3D interface class and node (and later, nodelet to be added). This is very similar to the K79(-2D) class and node.
• Merge branch \'nick-fixK79Parsing\'
• Merge branch \'master\' of https://github.com/AinsteinAI/ainstein_radar
• Refactor tracking filter to use dynamic reconfigure Refactored the radar target tracking filter to use dynamically reconfigurable parameters both for high-level target tracking and low-level individual Kalman Filters (all of which share the same per-target KF parameters).
• Add native PCL point type for radar data conversion Added a new PCL point type based on pcl::PointXYZ with additional fields specific to radar data. This allows for visualizing targets based on range, speed, snr etc in RViz by color as can normally be done for x,y,z coloring which is useful for debugging. More importantly, this exposes radar-specific data to all PCL library functionality, for instance making easy to use any of the basic PCL filters wrapped in pcl_ros (http://wiki.ros.org/pcl_ros#ROS_nodelets) without adding any new implementations (though this will require nodelets which will be added next). This opens the posibility to do much more advanced filtering on radar data going forward.
• Fix parsing of K79 data using casts Added static_cast\'s to fix the parsing of K79 data from the char buffer
• Update usharp3d testing launch, fix bug in driver Updated the usharp3d testing launch file after outdoor testing. Fixed a time bug in the usharp3d python driver by switching to Time(0) which gets the latest message regardless of stamped time.
• Add support for new uSharp3D version New uSharp3D version firmware was modified to change the data format bsed on feedback given to Beijing. The python ROS node was modified to support old and new firmware using a ROS param to specify which.
• Add device and frame ID params to uSharp3D node The uSharp3D node is still the rough python node (yet to be re-written in c++) but optional frame_id and device_id parameters were added for modularity. The defaults are \"map\" and \"/dev/ttyUSB0\", respectively.
• Modify and tune tracking filter, add test launch Modified the tracking filter to only use a measurement for one filter, rather than allowing multiple filters to use the same target. This was necessary because there were cases in which multiple filters would split off from one due to an inconsistent but valid target, and then these duplicates would be corrected by future, more accurate measurements and ALL remain in the list because they were all able to update with the same measurement. Tuned the tracking filter on single and two target data from outdoors testing of uSharp3D. This radar has particularly \"jumpy\" returns which require setting the confidence level and filter timeout lower to keep filters alive through measurement gaps. The process noise was decreased to keep the measurement covariance lower and prevent using bad targets. The measurement noise was increased to \"smooth\" jumps due to spotty targets, resulting in more of a low-pass tracking. Note: tuning was done by first filtering out distance targets using the range filter (>10m).
• Add rosconsole configuration file for debugging Added a config/ folder containing a rosconsole configuration file which enables printing DEBUG-level messages. To enable this from a launch file, load the rosconsole config into the environment with: <env name=\"ROSCONSOLE_CONFIG_FILE\" value=\"\$(find ainstein_radar_drivers)/config/debug_rosconsole.conf\"/>
• Add range filter for RadarTargetArray Added a range filter to remove targets from a RadarTargetArray which are outside specified parameter bounds for min/max range. Added a node and nodelet based on this filter, with the intention being that other filters (for example, speed filter, angle filter, etc) can be run as nodelets and stacked for easy preprocessing of raw detections.
• Rename defines, add val gate thresh param Renamed the noise-related KF parameter defines to STDEV to make it clear that these are standard deviations, not variances (they get squared in the noise matrices). Rescoped filter parameters to a /filter namespace and added a validation gate threshold to the filter which is set by the user based on desired confidence level from a Chi Squared distribution with 4 DoFs (measurement dimension). These can be looked up in a table for now, eg the table \"Lower-tail critical values of chi-square distribution with