Multi-Robot Asynchronous Planning and Execution for Cooperative Assembly
Philip Huang*,
Ruixuan Liu*,
Shobhit Aggarwal,
Changliu Liu,
Jiaoyang Li
Carnegie Mellon University
RSS 2025
This is the code repository for APEX-MR, our planning and execution framework for dual-arm LEGO assembly.
Gurobi (optional):
If you would like to search over all possible LEGO grasp/support poses (i.e.
Build the docker image and run it inside docker
cd docker && bash build.sh
If you are not using the docker file, the following setup has been tested on Ubuntu 20.04 with ROS Noetic. You may need to install some system dependencies
Once you have download ros, and other system deps, under your catkin_ws/src
, download the robot model to your workspace
- gp4 digital twin Checkout to the
apexmr-release
tag!
Under the catkin_ws/src
workspace, doanload the code. Then use catkin build
to compile the code.
To generate the task assignment for lego task, run
roslaunch apex_mr lego_assign.launch task:=cliff
You should see a message saying lego assignment success.
To compute the motion plan for lego assembly and build the corresponding TPG for asynchronous execution, run
roslaunch apex_mr lego.launch task:=cliff
You should see a lego planning success
vmax:=X
to set maximum velocity scale, default X = 1
adg_shortcut_time:=X
to set the time for TPG shortcutting, default X = 1s
sync_plan:=true
to run the sync planning baseline
sync_shortcut_time:=X
to set the time for RRT-Connect,shortcutting, default X = 0.1s
Target LEGO assembly are specified under config/lego_tasks/assembly_tasks
.
Description of the LEGO assembly plate are specified under config/env_setup/assembly_tasks
We provide a script to run all 9 LEGO assemblies from the paper in simulation
python3 scripts.py/benchmark.py
Note that the exact results may be slightly different from the numbers reported in our paper.
If you find this repository useful for your research, please cite the following work.
@inproceedings{huang2025apexmr,
title = {APEX-MR: Multi-Robot Asynchronous Planning and Execution for Cooperative Assembly},
author = {Huang, Philip and Liu, Ruixuan and Aggarwal, Shobhit and Liu, Changliu and Li, Jiaoyang},
year = {2025},
info = {https://intelligent-control-lab.github.io/APEX-MR/},
booktitle = {Robotics: Science and Systems},
url = {https://arxiv.org/abs/2503.15836}
}