COMP0132 MSc Robotics and Computation Project

Integrating Road Marking Extraction From LIDAR Intensity into HDL Graph SLAM Framework

Road Marking Extraction Algorithm is a project focused on Road Marking Extraction from LIDAR Intensity Data using a Intensity Thresholding Method

Installation

To run this package, please

Install the hdl_graph_slam package utilised for building the map and upon which the road marking extraction is integrated. Please follow for detail on how to install this package hdl_graph_slam
Install all dependencies - Libraries and ROS packages
- Libraries - OpenMP, PCL, g2o, suiteparse, geodesy, nmea_msgs. Please follow Librariesfor instruction.
- Packages - geodesy, nmea_msgs, pcl_ros, ndt_omp, fast_gicp, glog_catkin Please follow packages for instruction .
Install the CloudCompare software, to visualise the map built using this package and also compare with the groundtruth map please follow: CloudCompare.

Install this package by cloning the repository into the src folder of your workspace:

cd ../your_ws

git clone https://github.com/abdulbaasitt/road_markings.git

Build and source the workspace

catkin build

source devel/setup.bash

In another Terminal, launch the package using this command:

roslaunch road_markings hdl_gs_new.launch

Launch RVIZ and play the bag files contained in the dataset on OneDrive:

roscd hdl_graph_slam/rviz
rviz -d hdl_graph_slam.rviz
rosbag play --clock PSA_APM_8380_AIDE_TB2_TB3L1_WRK_X6_2021-09-03-11-07-08_0.bag

To build the a full map of the Area in the dataset. Please Run all the bags files in OneDrive and save the map built on RVIZ using:

 rosrun pcl_ros pointcloud_to_pcd input:=/hdl_graph_slam/map_points

Recommendations

It is highly recommended to install AIOS to make the RVIZ visualization and running of the bag files easier.

To Install AIOS, Please follow AIOS

Datasets

Find the dataset(ROSbag files) used for testing this package OneDrive

Note: to access this dataset. Kindly contact the author. The dataset is the IP of AIDrivers Ltd where this project was carried and requires explicit authorisation for use from AIDrivers LTD.

Evaluation

To Evaluate, Segment the ground truth map to include on the area contained in the bag files from the Dataset,

Align the map obtained using this package with the ground truth map using cloudcompare

Save the two maps as Images and use the Evaluation to compare.

Licence

This project is authored by Abdulbaasit Sanusi and is licensed under the MIT License.

Enquires

Please contact the author via the emails providedbelow for any enquiries recagarding the running of this package: Email: abdulbaasitsanusi@gmail.com; abdulbaasit.sanusi.21@ucl.ac.uk

abdulbaasitt/COMP0132_RHWT4