ahanjaya/lfh

Learning from Hallucination

Learning from Hallucination (LfH) - Proof of Concept (PoC)

Description

• This repository contains packages for simulating Learning from Hallucination paper. Using the same robot as BARN Challenge.

Getting started

ClearPath Jackal for noetic

• Install ROS Noetic, make sure you also installed Gazebo and RViz.
• Add Clearpath custom metapackages here.
• Follow this instruction.
• Install all jackal packages.

  sudo apt install ros-noetic-jackal-*
  

• Clone this repo and run catkin_make in the workspace. Remember to checkout the correct branch.

  mkdir ~/catkin_ws
  cd ~/catkin_ws 
  git clone https://github.com/ahanjaya/LfH.git src/
  cd ~/catkin_ws
  catkin_make -j16
  

1. Collect training data.

• Terminal

  roslaunch lfh_data collect_data.launch
  

2. Convert training data.

• Convert .bag to ros2 format. According to rosbags repository

  rosbags-convert <output_name>.bag

• Terminal

  roslaunch lfh_data convert_data.launch

3. Train the model.

• Terminal

  roslaunch lfh_data train_data.launch

Setting RVIZ

• Copy rviz setting lfh_navigation.rviz into /opt/ros/noetic/share/jackal_viz/rviz.

  sudo cp ~/catkin_ws/src/lfh_utils/rviz/lfh_navigation.rviz /opt/ros/noetic/share/jackal_viz/rviz

Run the simulation and navigation

• Terminal

  roslaunch lfh_inference bringup_lfh.launch