============================================================ == Object Detection using LiDAR point cloud ============================================================ Object detection, tracking, and classification using LiDAR point cloud - Current implementation: ROS with Python - Classifier is not optimized yet ============================================================ == What's included ============================================================ object_detection +-- include | +-- Eigen | +-- Tracker.h | +-- kalman_filter.h +-- launch | +-- object_detection.launch +-- rviz_cfg | +-- obj_detect.rviz +-- scripts | +-- classification_svm.py | +-- connected.py | +-- features_extraction.py | +-- object_detection_node.py | +-- occupancy_grid.py | +-- scan_conversion.py | +-- segmentation.py | +-- trained_classifier_4classes.pkl +-- src | +-- Tracker.cpp | +-- kalman_filter.cpp | +-- tracking_node.cpp +-- CMakeLists.txt +-- package.xml +-- readme.txt (this file) +-- requirements_python36.txt ============================================================ == Getting Started ============================================================ Prerequisites: -Ubuntu 16.04 with ROS Kinetic (Installation instruction: http://wiki.ros.org/kinetic/Installation/Ubuntu) -Catkin workspace (how to create catkin workspace: http://wiki.ros.org/catkin/Tutorials/create_a_workspace) -Python 3.6 (tested with Python 3.6, but 3.5 should work) -Phtyon libraries: numpy, scipy, scikit-learn, etc. (requirements file (requirements_python36.txt) is provided just in case; create your virtual environment, and run: pip install -r requirements_python36.txt) How to build with catkin: $ unzip ros_object_detection.zip -d ~/catkin_ws/src/ $ cd ~/catkin_ws $ catkin_make $ source ~/catkin_ws/devel/setup.bash Python scripts are under ~/catkin_ws/src/object_detection/scripts ============================================================ == How to run ============================================================ $ roslaunch object_detection object_detection.launch (This launches three separete programs, which can be run in three separete terminals as follows: $ roscore $ rosrun object_detection object_detection_node.py $ rosrun object_detection tracking_node $ rosrun rviz rviz -d ~/catkin_ws/src/object_detection/rviz_cfg/obj_detect.rviz ) In second terminal, play sample rosbag file including LiDAR data (VLP-32C): $ rosbag play sample_rosbag.bag Note: Please find the links below for ROSBAG with Velodyne LiDAR Data and Camera https://app.box.com/s/nueax6f2ylcspf796xk13xnoxb2zra71 https://velodyne-my.sharepoint.com/:f:/p/algorithmteam/EruWWshc4gtCqxHYCNYy4JEBq3GGg8QM8RpPssV6AbxjsA?e=qrbAeq * 2017-10-01-19-54-57_Velodyne-VLP-16-Data.pcap file and _2017-12-12-21-58-53_CESDemo.bag file are VLP-32 bag file. * 2017-10-16-13-23-15_Alameda_Marina.pcap file is VLP-16 bag file.