v3.7.x

v3.7.x

• Add support for the new Neural Depth mode
• Add support for head detection model
• Add support for sport-related object class
• Add support for X_MEDIUM neural network models
• Enable AI for ZED Mini
• Add new <zed>_base_link frame on the base of the camera to easily handle camera positioning on robots. Thx @civerachb-cpr
• Improve URDF by adding 3° slope for ZED and ZED2, X-offset for optical frames to correctly match the CMOS sensors position on the PCB, X-offset for mounting screw on ZED2i
• Add zed_macro.urdf.xacro to be included by other xacro file to easily integrate ZED cameras in the robot descriptions. See PR #771 for details. Thx @civerachb-cpr
• New parameter save_area_memory_db_on_exit to force Area Memory saving when the node is closed and Area Memory is enabled and valid.
• Add service save_Area_map to trigger an Area Memory saving.
• New tool function to transform a relative path to absolute.
• Enabled static IMU TF broadcasting even it publish_tf is set to false, making the two options independent. Thx to @bjsowa
• Moved the zed_interfaces folder in the new zed-ros-interfaces repository. The new repository is useful to receive the topics from a ZED node on system where the zed-ros-wrapper repository cannot be fully installed, i.e. systems without CUDA support. For this repository nothing changes because the zed_interfaces folder is replaced by the zed-ros-interfaces git submodule to automatically satisfy all the dependencies.
• Fix sensors topics pubblication for ZED2i. The support for the new camera was not complete
• Fix sensor_msgs type for depth image in OpenNI mode, from sensor_msgs::image_encodings::mono16 to sensor_msgs::image_encodings::TYPE_16UC1. Depth image in OpenNI mode is now compatible with the nodelet depthimage_to_laserscan