3d_projection

This repo contains ros package for publishing 3D landmark pose, label and ID for robotX boat.

Input topics:

bbox_topic = "/darknet_ros/bounding_boxes"
pcloud_topic = "/wamv/sensors/lidars/lidar_wamv/points"
localization_topic = "/wamv/robot_localization/odometry/filtered"

bbox_topic contains information of object bounding boxes and their labels.
pcloud_topic is for LiDAR point cloud.
localization_topic is given by SLAM team, it containes boat poses in map frame.

Output topics:

landmarks_topic = "/landmark_detection/landmarks"
landmark_lla_topic = "/vrx/perception/landmark"

landmarks_topic publishes the landmark labels and landmark positions in base frame.
landmark_lla_topic publishes the landmark labels and landmark positions in LLA coordinate (which is required by the RobotX tasks).

To Run the package

In catkin_ws/src

git clone https://github.com/MichiganRobotX/3d_projection.git
cd 3d_projection/landmark_detection
mv darknet_ros_msgs ../..
cd ../../..
catkin_make
source devel/setup.bash
rosrun landmark_detection landmark_detector.py

To Run the package with darknet_ros together:

In catkin_ws/src

roslaunch landmark_detection perception.launch

If shows the error ‘can not launch....’, run the following command lines first.

Chmod +x /path/to/landmark_detector.py
Chmod +x /path/to/perception.launch