Environment Aware and Safe Tracking (EAST)

This repository contains code for the EAST project as supplementary material for the paper submission at IJRR. ArXiv version is now available at here.

This code demonstrate the capability of EAST algorithm in various environments for safe and environemnt aware navigation in challenging environments.

This code is primarily developed by Zhichao Li , Yinzhuang Yi and Zhuolin Niu from Exsitential Robototics Laboratory led by Professor Nikolay Atanasov at UCSD.

Dependencies

The code is tested on Ubuntu 20.04 LTS + ROS Noetic

ROS libaries

• JACKAL sudo apt install ros-noetic-jackal-*
• HECTOR_MAPPING sudo apt install ros-noetic-hector-mapping
• RQT_MULTIPLOT sudo apt install ros-noetic-rqt-multiplot
• PYBIND11_CATKIN sudo apt install ros-noetic-pybind11-catkin

Non-ROS libaries

• Eigen3 sudo apt install -y libeigen3-dev
• Boost sudo apt install-y libboost-dev
• OPENCV sudo apt install libopencv-dev python3-opencv
• CVXPY pip install cvxpy with your favourite QCQP solver (MOSEK, Gurobi, etc.)
• CGAL sudo apt install libcgal-dev

Complication

Download this repo, create working space and use catkin build tool to build it just as other ROS packages. Don't forget to source the environment afterward. All non-official packages have been included.

Be patient, Gazebo starts slow for initial installation. Wait until the robot shows up in Rviz or enable Gazebo GUI in launch file.

Demonstration

Evaluate EAST in simulation at Jackal Race World.

Run the following script in ros-sourced terminal first,

roslaunch ref_gvn_ros ref_gvn_cone.launch

and click goal point within map using RViz Nav2D button.

Evaluate EAST in simulation at U-shape to see the effect of SDDM boost. By running

Run the following script in ros-sourced terminal first,

• sddm boost enabled

  roslaunch ref_gvn_ros ushape_sddm_on.launch

• sddm boost disabled

  roslaunch ref_gvn_ros ushape_sddm_off.launch

and click any point on the map using RViz Nav2D button.

Evaluate EAST in simulation at Maze.

Run the following script in ros-sourced terminal first,

roslaunch ref_gvn_ros ref_gvn_cone_maze.launch

and click goal at (6, 6) using RViz Nav2D button on the map. This experiment might experience de-localization problem randomly. Feel free to test our algorithm in other maps/worlds.

Evaluate EAST in simulation at Jackal Race World with virtual moving obstacles.

Run the following script in ros-sourced terminal first,

roslaunch ref_gvn_ros act_refgvn_virtual_movobs.launch

and click goal point within map using RViz Nav2D button.

Note that, there is no gurantee for all time safety, instead, our controller just tries its best to avoid moving obstacle. No action will be taken when goal is reached.

Citing this work

If you find the ideas and code implementation are useful for your own work, please check out the following paper and more related projects at my website.

• Environment Aware and Safe Tracking (EAST) [pdf]

    @article{li_east,
      author = {Li, Zhichao and Yi, Yinzhuang and Niu, Zhuolin and Atanasov, Nikolay},
      title = {EAST: Environment Aware Safe Tracking using Planning and Control Co-Design},
      journal = {arXiv preprint arXiv:2310.01363},
      year = {2023}
    }