Be notified of new releases
Create your free GitHub account today to subscribe to this repository for new releases and build software alongside 28 million developers.Sign up
- There is a security fix. Please apply #1674 if you want to use old versions.
- Indigo is marked as deprecated, and will be removed in a future release (#1663).
- Obsolete packages and nodes will be removed in future releases (#1664).
- Object motion can be predicted by naive_motion_predict (#1616).
- Static objects in point cloud map can be filtered out by compare_map_filter (#1559).
- vision_beyond_track smoothly tracks image objects (#1473).
- L-shape fitting for stably tracking cars (#1457).
- RoboSense LiDAR driver (#1500).
- Selectable topics in Runtime Manager (#1596).
- data_rate_checker_plugin visualizes topic rate (#1491).
- Nodes and topics can be monitored by diagnostic_lib (#1516, #1528, #1527).
- Customizable classification name in darknet YOLO (#1535).
- Cross build for driveworks no longer requires the full NVIDIA SDK (#1534).
- Docker image is built from local repo instead of cloning GitHub (#1635).
- The settings of ImageViewerPlugin are saved with the RViz configuration (#1511).
- Fixed an issue where docker image can not be built (#1536).
- Fixed an issue where ndt_mapping will fail when lidar is mounted in a different direction from base_link (#1472).
- Fixed an issue where euclidean cluster dies if not enough input points (#1606).
- Fixed an issue where euclidean cluster doesn't work on CPU w/o multiple thres (#1606).
- Fixed an issue where catkin runs multiple times in catkin_make_release (#1620).
- Fixed the setup instructions of SSD (#1613).
- Fixed an issue where lidar_imm_ukf_pda_track is not connected to the other nodes (#1582).
- Fixed the compilation error of darknet on Indigo (#1608).
- Fixed an issue where Topic Frequency in DataRateCheckerPlugin is invisible depending on window size (#1515).
- Fixed an issue where Point Size of ImageViewrPlugin is not saved with RViz configuration (#1515).
- Fixed an issue where the intrinsic calibration node outputs an anti-aliasing error (#1581).
- Fixed the yaml compatibility issue in autoware_camera_calibration (#1622).
- Fixed an issue where the coodinate frame of camera_info on frame_id is not considered (#1565).
- Results of vision detector and lidar detector can be combined by range_vision_fusion (#1419).
- Colors of vision data can be projected into lidar data by pixel_cloud_fusion (#1297).
- Fake perception generating fake object and pointcloud message (#1439).
- Cross build for NVIDIA DRIVE PX2 (#1447) and generic aarch64 (#1433).
The build time for NVIDIA DRIVE PX2 is shortened dramatically.
- Docker for 96boards (#1356).
- Drivers of Sick LD-MRS lidar (#1287).
- Grid divider of Point Cloud Map (#1271).
- Convenient autoware_bag_tools that rename frame IDs (#1307) and extract GNSS nmea data (#1345).
- Using TF generator when tf.launch is not given (#1239).
- Pure pursuit publishes deviation of current position (#1357).
- Can limit maximum scan range of ndt_mapping (#1412).
- Euclidean cluster classifies pointcloud more precisely (#1426).
- Up to 8 lidars are supported by concat filter (#1255).
- Speed-up Points to Image processing (#1286).
- Brush up the internal structure of OpenPlanner (#1400).
- Fixed an issue where multiple cameras publishes the same frame id (#1296, #1313).
- Fixed ndt_matching and ndt_mapping with GPU of Drive PX2 and Jetson TX2 (#1340).
- Fixed an issue where /can_velocity and /vehicle/odom are not generated from /vehicle_status (#1351).
- Fixed an issue where transform from map to base_link is not published in ndt_mapping (#1376).
- Fixed an issue where the app button of lidar_kf_contour_track and lidar_imm_ukf_pda_track do not work (#1436).
- Kanade – Lucas – Tomasi (KLT) Tracker does not work on Indigo (#1415).
- Fix the multi cam support of Traffic Light recognition feat_proj node.
- Add support to multi traffic signal located in the same pole.
- Add support to ueye from IDS Imaging cameras using the ueye ros package.
- Update Hokuyo 2D driver.
- Add multi LiDAR calibration tool.
- Add Camera-LiDAR calibration tool.
- Enable Matlab engine for chessboard detection.
- Update velodyne drivers and new VLP32C driver.
- Add support to AVT cameras usin the Vimba SDK.
- Add traffic light recognizer based on MxNet framework.
- Add Yolo v3 detector.
- Add accleration to ctrl_cmd.
- Add vehicle status node/topic.
- Fix compile error on decision_maker.
- Fix velocity planning at end of waypoint.
- Updated README, Added FAQ, Travis-CI status, new nodes.
- Add NDT matching safe monitor.
- Update Docker README.md.
- Add multiple objects trakking algorithms by IMM-UKF-PDA tracker.
- Add velocity re-planning tool for waypoint.
- Update concat filter to increase the supported number of lidars from 2 to 8.
- Rename and restructure packages
- Add GPU NDT Matching
- Add controllable vehicle from Autoware. The following cars will be controllable:
- Yamaha Motor Powered Products Co., Ltd. G30Es Li
- Autonomous-stuff, PACMod equipped
- Add new Rviz models.
- Add docker file for the ubuntu 16.04 with ROS Kinetic
- Add Ray Ground Filter
- Enhance DecisionMaker, it can control vehicle according to rule based.
- Enable to perform split loading of maps in NDT in real time.
- Add Ray ground filter. it can be filtering ground when using multiple velodyne.
- Add to support multi traffic signal
- Deprecated WebRTM
- Fix some bugs
- rviz window position
- NDT Matching does not work when lidar rotates.
- When "use_gpu" of ndt_matching (also ndt_mapping) is true,
As the map size increases, the CUDA memory is insufficient and the program drops
- When "use_fast_pcl" of ndt_matching (also ndt_mapping) is true, if load a large map, use too huge memory.
- Add Web-UI
- Add Decision Maker [WIP]
- Add GarminGPS support
- Add multiple cross-walk detector
- Add ENet image segmenter
- Add aarch64 support and NVIDIA DrivePX2 installation manual
- Add simulated obstacle creator for astar planning/dp_planner
- Fix docker files, Update nvidia-docker version.
- Fix some issues.