jrenaf / ltl_multi_agent

Reactive Task Allocation and Planning for Quadrupedal and Wheeled Robot Teaming

Overview

This repo contains all the packages for multi-agent task allocation and planning based on Linear Temporal Logic (LTL). Each component is maintained individually. To run the simulation, ROS melodic and gazebo11 are suggested.

Packages

This metapackage is composed of the following packages.

• ltl_planning_core: A metapackage in charge of multi-robot task allocation and planning.

• ltl_automation_a1: Behavior tree nodes for online task execution. Quadruped and wheeled robots are both included.

• quadruped_sim: Essential components for quadruped including Gazebo simulation, navigation, and low-level control. Make sure you have Cheetah-Software installed before building the package.

• motion_capture_simulator:

• ubtna_gazebo_pkg: Essential components for wheeled robots from UBTECH including Gazebo simulation, navigation

• Groot (optional): GUI for behavior tree

Installation

Before building the whole workspace, make sure quadruped_ctrl package is built successfully which is inside quadruped_sim folder.

All packages shall be located in the same catkin workspace. To build the packages, follow:

$ cd catkin_ws/src
$ git clone --recurse-submodules https://github.com/GTLIDAR/ltl_multi_agent
$ cd ..
$ rosdep install --from-paths src --ignore-src -r -y
$ catkin build
$ source devel/setup.bash

Usage

Run a case study that involves a quadruped A1, a delivery robot DR and a walk training robot Wassi (DR and Wassi models are proprietary and presented by a simple model):

In terminal 1, start gazebo simulation and BT nodes for each robot. Try rerunning the launch file if models are not spawned correctly (to be fixed).

$ roslaunch ltl_automaton_planner study_2.launch

In terminal 2, start quadruped controller:

$ rosrun quadruped_ctrl a1_servo /cmd_vel:=/a1_gazebo/cmd_vel

In terminal 3, start LTL planner if quadruped stands up and laser scans are received in Rviz:

$ rosrun ltl_automaton_planner robot_planner_node_exp

Reference Citation

To cite this work:

@inproceedings{zhou2022reactive,
  title={Reactive Task Allocation and Planning for Quadrupedal and Wheeled Robot Teaming},
  author={Zhou, Ziyi and Lee, Dong Jae and Yoshinaga, Yuki and Balakirsky, Stephen and Guo, Dejun and Zhao, Ye},
  booktitle={2022 IEEE 18th International Conference on Automation Science and Engineering (CASE)},
  pages={2110--2117},
  year={2022},
  organization={IEEE}
}