ROS-AD-SLAM-Evaluation

How to setup Carla Environment to replicate the test results

Requirements

• Ubuntu 18.04.
• 16gb RAM.
• Vulkan enabled GPU is preferred , Follow the LINK to check your GPU compatibility.
• Unreal Engine 4.24 , for build instructions follow LINK.
• Carla 0.9.8, for build instructions follow LINK.
• ROS Melodic , for install instructions follow LINK.

Setup the ROS Environment

• Navigate to carla build directory , you should find *.egg file in carla/Dist/CARLA_Shipping_0.9.8-***********/LinuxNoEditor/PythonAPI/carla/dist/ folder, which to be exported to python environment to run further simulations. Make sure you export appropiate egg file based on your preferred python version.
• Install SLAM Nodes - Install Gmapping by sudo apt-get install ros-melodic-slam-gmapping. - Install Hector SLAM by sudo apt-get install ros-melodic-hector-slam. - Install RTAB-Map by sudo apt-get install ros-kinetic-rtabmap-ros. - Karto SLAM needs to be build from the source , follow standard ROS package installation procedure. The Kartoslam Github repo can be found LINK.
• Clone the SLAM packages https://github.com/prithvisekhar/ROS-AD-SLAM-Evaluation.git
• Copy the ros-bridge contents to a ros workspace src folder and build the workspace and source the workspace before running the roscore.

Running the Simulator

• Run Carla simulator by navigating to carla/Dist/CARLA_Shipping_0.9.8-***********/LinuxNoEditor/ and running ./CarlaUE4.sh if you are using opengl then you should pass -opengl as configuration along with the carla launch command. it should open the simulator (Refer the below image as reference). For More info on graphics support follow this LINK
• Run the Carla ROS Client roslaunch carla_ros_bridge carla_ros_bridge_with_example_ego_vehicle.launch , Refer the below image as reference
  • Run the Gmapping Node roslaunch carlabg Gmapping_Carla.launch , for configuration option see ros-bridge/carlabg/launch folder. After mapping session is over save the map using map_saver tool rosrun map_server map_saver -f /path to your file.
  • Run the Hector SLAM Node roslaunch carlabg Hectormapping_Carla.launch. Follow the same map saver command to save the map for Hector SLAM.
  • Run the Karto SLAM Node roslaunch carlabg KartoSLAM2.launch. Follow the same map saver command but this time pass the map frame as map2.
  • Launch the RTAB-Map Node roslaunch carlabg RtabMapping_Carla.launch. RTAB Map can be directly saved from the RTAB map application.