Changelog for package humanoid_localization
0.4.1 (2016-09-05)
- catkinize the stack
- get code to compile under Hydro
- removes some trailing blanks and semicolon
- added localization Service to HumanoidLocalization (untested)
- Localization updates are now done based on the distance to the last localized pose (not summed distance)
- Homogeneous subsampling of laser points (PCL voxel grid) instead of angular
- Uniform sampling for depth camera data
- Motion model and noise can now be calibrated with a variance matrix. Some parameters are no longer used, please adjust accordingly:
motion_noise/x, motion_noise/y, motion_noise/yaw, num_sensor_beams
- Preventing TF_OLD_DATA Warning when init from RVIZ
- refactoring/cleanup
- Added Timer to Localization and timerCallback now republishes latest tf transform
- added parameterization for odometry calibration in MotionModel
- Proper floor offset if z is constrained to supporting surface by kinematics / odometry
- More loops parallelized (motion model)
- fixes:
- Fixed invalid conversion from 'octomap::AbstractOcTree*' to 'octomap::OcTree*'
- Add dependency to cmake_modules for FindEigen.cmake
- Fixed duplicate library export
- Fixed linking endpointmodel to humanoidlocalization in case dynamicEDT3D is available
- Fix missing OpenMP compilation flags (from catkinization)
- Fixing error with constrainMotion flags and optimized-away assertion
- Fixed point cloud callback crashing when ground filtering and no ground visible
- Bugfix for OctoMap in Groovy
- Height lookup for endpoint model fixed
- Fixed raycasting height integration on non-zero ground and raycast max_range
- bug fix: removed (unnecessary) feature in HumanoidLocalization
- Fixed compilation with ROS groovy (tf types)
- fixed localization for point clouds
- Contributors: Armin Hornung, Daniel Maier, Pramuditha Aravinda, Stefan Osswald, Vincent Rabaud, enriquefernandez
0.4.0 (2013-01-10)
- indentation / code style
- added point cloud support (with ground filter) to humanoid_localization
- Added option to constrain motion according to odometry (don't estimate z / RP)
- cleanup, init localization from ground truth via tf lookup
- Fixed odometry lookup error (wrong frame) and missing odom data
- restructuring of humanoid_localization started
- Parameter cleanup / renaming
- Better height measurement in raycasting model
- humanoid_localization now uses sensor_msgs::Imu instead of nao_msgs
- EndpointModel is now optional, depends on dynamicEDT3D
- humanoid_localization added to humanoid_navigation, currently supports 6D localization from laser, odometry and IMU
- Contributors: Armin Hornung, Daniel Maier