autoware.learn

Environment :

• ubuntu 18.04.1 LTS
• amd64
• cuda 10.0 (RTX3060)
• ros-melodic-desktop-full

installation :

• autoware.ai generic DOCKER -v : 1.13.0

First compilation

mkdir -p autoware
cd autoware
git clone
# change the folder name from autoware.learn to src

AUTOWARE_COMPILE_WITH_CUDA=1 colcon build --cmake-args -DCMAKE_BUILD_TYPE=Release

source install/setup.bash
echo export LIBGL_ALWAYS_SOFTWARE=1 >> .bashrc ## for rviz

To compile a single (ros)package

AUTOWARE_COMPILE_WITH_CUDA=1 colcon build --cmake-args -DCMAKE_BUILD_TYPE=Release --packages-select <package_name> 

setup plugins

download file : libActorCollisionsPlugin.so into /usr/lib/x86_64-linux-gnu/gazebo-9/plugins/

sudo apt update
sudo apt-get install gedit
sudo apt-get install ros-melodic-urdf-tutorial
sudo apt-get install ros-melodic-rqt-robot-steering
sudo apt-get install ros-melodic-hector-gazebo-plugins

Rosbag set start time

rosbag play -s n <bagname>

Gazebo simulation with rviz

roslaunch autoware_quickstart_example mini_map.launch
roslaunch autoware_quickstart_example mini_localization.launch
roslaunch autoware_quickstart_example mini_sil_env.launch

Euclidean Cluster

# run Gazebo simulation with rviz
roslaunch lidar_euclidean_cluster_detect lidar_euclidean_cluster_detect.launch

kf Contour Tracker

# require Euclidean Cluster realtime results
roslaunch lidar_kf_contour_track lidar_kf_contour_track.launch

Mapping with ndt-mapping

Note: ndt-mapping with demo rosbag

• ndt-mapping with simcity

Vector Map Builder by autoware.tools

Official Demo