"/drone/nagvation/pos", PoseStamped, come from control node and refers to the target position
"/auto_mode/status", Bool, come from control node and refers to whether turns to auto mode
"/detection_result/number_of_obstacle", Int32, come from detection node and refers to number of obstacle
"/detection_result/number_of_human", Int32, come from detection node and refers to number of human
"/detection_result/number_of_injury", Int32, come from detection node and refers to number of injury
"/extract_indices/output", Int32, come from detection node and refers to the pointcloud that processed by filter
"/drone/input_postion/pose", PoseStamped, send to control node and refers to the position that the drone should be go
"/desired_path/position", MarkerArray, for observe purpose and refers to the approximated obstacle centroid
"/desired_path/validation_position", MarkerArray, for observe purpose and refers to the point around the obstacle centroid
"/desired_path/local_marker", MarkerArray, for observe purpose and refers to the drone path
2021-8-26 upload environment.py which can allow the obstacle to place into the game field randomly
2021-8-27 upload motion.py which allow to maintain particular height and proivde rgb, depth, infra1 and infra2 drone view
2021-8-30 upload empty_world.world, calibrate_world.world and camera_calibration.launch to prepare the camera calibration environment
2021-8-31 update motion.py to provide distance data via depth message
2021-9-4 update motion.py to provide dynamic filter for the cropped region pointcloud and write the record the point cloud XYZ into csv file
2021-9-4 upload plot_csv.py which can povide the visual animate 3d plot from csv file data
2021-9-5 update camera_calibration.launch, spawn.launch and field_realsense.launch to visualize the drone model in rviz and correct the right tf between the drone and the world
2021-9-6 upload camera_capture.py which allow to press keyboard button to capture the camera frame and save it into specific folder
2021-9-7 upload camera_image_callback.py which include camera callback class specially for return D435i camera data
2021-9-10 upload test_depth_distance.py which can detect the bounding box distance and show the marker on the image window
2021-9-10 upload test_attention_clipper.launch which can allow input Bounding Box topic to change the point cloud ROI size
2021-9-10 upload attention_pose_set.py which can publish the input Bounding Box topic for test_attention_clipper.launch to change the Bounding Box size
2021-9-11 upload test_2d_to_3d.world, test_2d_to_3d.launch and box (0.5m) model for testing 2d_coordinate_to_3d_coordinate.py
2021-9-17 upload publish_marker.py which provide that the 2d rgb or depth image point to the 3d position rely on the below matrix
[ 2d point x] [fx 0 u 0] [3d point x]
depth distance to the 3d point [ 2d point y] = [0 fy v 0] [3d point y]
[ 1 ] [0 0 1 0] [depth distance to the 3d point]
[1]
the 3d point in rviz have a little bit error, it should be the calculated 3d point is based on the world frame as origin instead of d435 camera
After testing, it should be not a big problem as the point will set in front of the trunk which means the front part of the trunk will be consider
2021-9-19 upload yolov4-tiny folder which responsible for detecting the tree and dynamic_filter.launch for launch the clipped point cloud
2021-9-19 upload dynamic_filter.rviz for visualize the clipped point cloud performance
2021-10-19 upload SETUP.md for PX4 Gazebo with ROS Wrapper environment setup
2021-10-27 update SETUP.md and complete the PX4 Gazebo with ROS Wrapper environment setup
2021-10-29 successfully install realsense D435i into drone model and all the camera topic seems work normally, start to study VINS-FUSION for localization component
2021-11-8 finish the vins-fusion testing in Gazebo environment and start to study octomap
2021-11-10 finish the testing for vins-fusion part and octomap with downsampled point cloud in Gazebo, start to implement the object detection into the Gazebo
2021-11-11 update SETUP.md with VINS-fusion installation, octomap installation and QGC installation
2021-11-12 update SETUP.md with nvidia-driver, CUDA, cuDNN and Pytorch installation
2021-12-17 upload realworld_dynamic_filter_testing folder which able to find obstacle and people attention point cloud by dynamic filter perfectly
2022-1-11 update SETUP.md for ROS noetic version software and confirm that noetic version default python is Python3
2022-1-22 update SETUP.md with remote Jetson, Vrpn installation
2022-1-22 successfully convert the moving.py into moving_with_auto_tf.py which can change the transform between camera frame to the world frame
2022-1-25 NED or ENU system need to be confirm, the futher study on https://blog.csdn.net/qq_33641919/article/details/101003978
2022-2-11 Updated moving_with_static_obstacle_v8.py which included k-means cluster to find the cluster pointcloud size and automatically update the path with best current solution
2022-2-21 Steve: added note to setup jetson (SetupJetson.md)
2022-2-23 Steve: added note to setup jetson to boot from SSD refer readme SetupJetsonSSD.md
2022-3-3 Upload yolov5 in Pytorch which speed up with TensorRT
2022-3-6 Finish YOLOv4-tiny in Darknet which speed up with TensorRT, specifiy details in others repo
test_attention_clipper.launch will automatically map the input Bounding Box into the point cloud, it seems to map the original point into d435_depth_optical_frame
attention_pose_set.py will set the bounding box center via pose.position but the bounding boxwill defined by dimensions
sudo apt-get install ros-melodic-geographic-msgs for the hector localization
/d435/depth/color/points is organized point cloud and datatype is XYZRGB