CTRMs: Learning to Construct Cooperative Timed Roadmaps for Multi-agent Path Planning in Continuous Spaces

Overview

CTRMs: Learning to Construct Cooperative Timed Roadmaps for Multi-agent Path Planning in Continuous Spaces

This is a repository for the following paper:

  • Keisuke Okumura, Ryo Yonetani, Mai Nishimura, Asako Kanezaki, "CTRMs: Learning to Construct Cooperative Timed Roadmaps for Multi-agent Path Planning in Continuous Spaces," AAMAS, 2022 [paper] [project page]

You need docker (≥v19) and docker-compose (≥v1.29) to implement this repo.

Demo

(generated by ./notebooks/gif.ipynb)

Getting Started

We explain the minimum structure. To reproduce the experiments, see here. The link also includes training data, benchmark instances, and trained models.

Step 1. Create Environment via Docker

  • locally build docker image
docker-compose build        # required time: around 30min~1h
  • run/enter image as a container
docker-compose up -d dev
docker-compose exec dev bash
  • ./.docker-compose.yaml also includes an example (dev-gpu) when NVIDIA Docker is available.
  • The image is based on pytorch/pytorch:1.8.1-cuda10.2-cudnn7-devel and installs CMake, OMPL, etc. Please check ./Dockerfile.
  • The initial setting mounts $PWD/../ctrm_data:/data to store generated demonstrations, models, and evaluation results. So, a new directory (ctrm_data) will be generated automatically next to the root directory.

Step 2. Play with CTRMs

We prepared the minimum example with Jupyter Lab. First, startup your Jupyter Lab:

jupyter lab --allow-root --ip=0.0.0.0

Then, access http://localhost:8888 via your browser and open ./notebooks/CTRM_demo.ipynb. The required token will appear at your terminal. You can see multi-agent path planning enhanced by CTRMs in an instance with 20-30 agents and a few obstacles.

In what follows, we explain how to generate new data, perform training, and evaluate the learned model.

Step 3. Data Generation

The following script generates MAPP demonstrations (instances and solutions).

cd /workspace/scripts
python create_data.py

You now have data in /data/demonstrations/xxxx-xx-xx_xx-xx-xx/ (in docker env), like the below.

The script uses hydra. You can create another data, e.g., with Conflict-based Search [1] (default: prioritized planning [2]).

python create_data.py planner=cbs

You can find details and explanations for all parameters with:

python create_data.py --help

Step 4. Model Training

python train.py datadir=/data/demonstrations/xxxx-xx-xx_xx-xx-xx

The trained model will be saved in /data/models/yyyy-yy-yy_yy-yy-yy (in docker env).

Step 5. Evaluation

python eval.py \
insdir=/data/demonstrations/xxxx-xx-xx_xx-xx-xx/test \
roadmap=ctrm \
roadmap.pred_basename=/data/models/yyyy-yy-yy_yy-yy-yy/best

The result will be saved in /data/exp/zzzz-zz-zz_zz-zz-zz.

Probably, the planning in all instances will fail. To obtain successful results, we need more data and more training than the default parameters as presented here. Such examples are shown here (experimental settings).

Notes

  • Analysis of the experiments are available in /workspace/notebooks (as Jupyter Notebooks).
  • ./tests uses pytest. Note that it is not comprehensive, rather it was used for the early phase of development.

Documents

A document for the console library is available, which is made by Sphinx.

  • create docs
cd docs; make html
  • To rebuild docs, perform the following before the above.
sphinx-apidoc -e -f -o ./docs ./src

Known Issues

  • Do not set format_input.fov_encoder.map_size larger than 250. We are aware of the issue with pybind11; data may not be transferred correctly.
  • We originally developed this repo for both 2D and 3D problem instances. Hence, most parts of the code can be extended in 3D cases, but it is not fully supported.
  • The current implementation does not rely on FCL (collision checker) since we identified several false-negative detection. As a result, we modeled whole agents and obstacles as circles in 2D spaces to detect collisions easily. However, it is not so hard to adapt other shapes like boxes when you use FCL.

Licence

This software is released under the MIT License, see LICENCE.

Citation

# arXiv version
@article{okumura2022ctrm,
  title={CTRMs: Learning to Construct Cooperative Timed Roadmaps for Multi-agent Path Planning in Continuous Spaces},
  author={Okumura, Keisuke and Yonetani, Ryo and Nishimura, Mai and Kanezaki, Asako},
  journal={arXiv preprint arXiv:2201.09467},
  year={2022}
}

Reference

  1. Sharon, G., Stern, R., Felner, A., & Sturtevant, N. R. (2015). Conflict-based search for optimal multi-agent pathfinding. Artificial Intelligence
  2. Silver, D. (2005). Cooperative pathfinding. Proc. AAAI Conf. on Artificial Intelligence and Interactive Digital Entertainment (AIIDE-05)
PaSST: Efficient Training of Audio Transformers with Patchout

PaSST: Efficient Training of Audio Transformers with Patchout This is the implementation for Efficient Training of Audio Transformers with Patchout Pa

165 Dec 26, 2022
PyTorch implementation of the paper Dynamic Data Augmentation with Gating Networks

Dynamic Data Augmentation with Gating Networks This is an official PyTorch implementation of the paper Dynamic Data Augmentation with Gating Networks

九州大学 ヒューマンインタフェース研究室 3 Oct 26, 2022
Convert BART models to ONNX with quantization. 3X reduction in size, and upto 3X boost in inference speed

fast-Bart Reduction of BART model size by 3X, and boost in inference speed up to 3X BART implementation of the fastT5 library (https://github.com/Ki6a

Siddharth Sharma 19 Dec 09, 2022
AI创造营 :Metaverse启动机之重构现世,结合PaddlePaddle 和 Wechaty 创造自己的聊天机器人

paddle-wechaty-Zodiac AI创造营 :Metaverse启动机之重构现世,结合PaddlePaddle 和 Wechaty 创造自己的聊天机器人 12星座若穿越科幻剧,会拥有什么超能力呢?快来迎接你的专属超能力吧! 现在很多年轻人都喜欢看科幻剧,像是复仇者系列,里面有很多英雄、超

105 Dec 22, 2022
Image augmentation library in Python for machine learning.

Augmentor is an image augmentation library in Python for machine learning. It aims to be a standalone library that is platform and framework independe

Marcus D. Bloice 4.8k Jan 07, 2023
UnpNet - Rethinking 3-D LiDAR Point Cloud Segmentation(IEEE TNNLS)

UnpNet Citation Please cite the following paper if you use this repository in your reseach. @article {PMID:34914599, Title = {Rethinking 3-D LiDAR Po

Shijie Li 4 Jul 15, 2022
We present a regularized self-labeling approach to improve the generalization and robustness properties of fine-tuning.

Overview This repository provides the implementation for the paper "Improved Regularization and Robustness for Fine-tuning in Neural Networks", which

NEU-StatsML-Research 21 Sep 08, 2022
Adversarial-autoencoders - Tensorflow implementation of Adversarial Autoencoders

Adversarial Autoencoders (AAE) Tensorflow implementation of Adversarial Autoencoders (ICLR 2016) Similar to variational autoencoder (VAE), AAE imposes

Qian Ge 236 Nov 13, 2022
Code and models for ICCV2021 paper "Robust Object Detection via Instance-Level Temporal Cycle Confusion".

Robust Object Detection via Instance-Level Temporal Cycle Confusion This repo contains the implementation of the ICCV 2021 paper, Robust Object Detect

Xin Wang 69 Oct 13, 2022
PyTorch implementation of Deformable Convolution

Deformable Convolutional Networks in PyTorch This repo is an implementation of Deformable Convolution. Ported from author's MXNet implementation. Buil

411 Dec 16, 2022
Implementation of "Distribution Alignment: A Unified Framework for Long-tail Visual Recognition"(CVPR 2021)

Implementation of "Distribution Alignment: A Unified Framework for Long-tail Visual Recognition"(CVPR 2021)

105 Nov 07, 2022
A Python library for differentiable optimal control on accelerators.

A Python library for differentiable optimal control on accelerators.

Google 80 Dec 21, 2022
[peer review] An Arbitrary Scale Super-Resolution Approach for 3D MR Images using Implicit Neural Representation

ArSSR This repository is the pytorch implementation of our manuscript "An Arbitrary Scale Super-Resolution Approach for 3-Dimensional Magnetic Resonan

Qing Wu 19 Dec 12, 2022
An architecture that makes any doodle realistic, in any specified style, using VQGAN, CLIP and some basic embedding arithmetics.

Sketch Simulator An architecture that makes any doodle realistic, in any specified style, using VQGAN, CLIP and some basic embedding arithmetics. See

12 Dec 18, 2022
The implementation of the lifelong infinite mixture model

Lifelong infinite mixture model 📋 This is the implementation of the Lifelong infinite mixture model 📋 Accepted by ICCV 2021 Title : Lifelong Infinit

Fei Ye 5 Oct 20, 2022
An official TensorFlow implementation of “CLCC: Contrastive Learning for Color Constancy” accepted at CVPR 2021.

CLCC: Contrastive Learning for Color Constancy (CVPR 2021) Yi-Chen Lo*, Chia-Che Chang*, Hsuan-Chao Chiu, Yu-Hao Huang, Chia-Ping Chen, Yu-Lin Chang,

Yi-Chen (Howard) Lo 58 Dec 17, 2022
An e-commerce company wants to segment its customers and determine marketing strategies according to these segments.

customer_segmentation_with_rfm Business Problem : An e-commerce company wants to

Buse Yıldırım 3 Jan 06, 2022
Empirical Study of Transformers for Source Code & A Simple Approach for Handling Out-of-Vocabulary Identifiers in Deep Learning for Source Code

Transformers for variable misuse, function naming and code completion tasks The official PyTorch implementation of: Empirical Study of Transformers fo

Bayesian Methods Research Group 56 Nov 15, 2022
Research on Event Accumulator Settings for Event-Based SLAM

Research on Event Accumulator Settings for Event-Based SLAM This is the source code for paper "Research on Event Accumulator Settings for Event-Based

Robin Shaun 26 Dec 21, 2022
Pytorch Code for "Medical Transformer: Gated Axial-Attention for Medical Image Segmentation"

Medical-Transformer Pytorch Code for the paper "Medical Transformer: Gated Axial-Attention for Medical Image Segmentation" About this repo: This repo

Jeya Maria Jose 615 Dec 25, 2022