This repository contains the official code of the Planning for Learning Object Properties (PLOP) algorithm, which has been presented at the 37th AAAI Conference on Artificial Intelligence (AAAI-2023 Main track), for details about the method please see the paper.
The following instructions have been tested on Ubuntu 20.04.
- Clone this repository
git clone https://github.com/LamannaLeonardo/PLOP.git
- Create a Python 3.9 virtual environment using conda or pip.
conda create -n plop python=3.9
- Activate the environment
conda activate plop
- Install PyTorch (tested with version 1.10.0)
conda install pytorch==1.10.0 torchvision==0.11.0 torchaudio==0.10.0 -c pytorch
- Install AI2THOR (tested with version 4.0.0)
pip install ai2thor==4.0.0
- Install the following dependencies
pip install matplotlib scipy ipython pandas tqdm seaborn urllib3 requests
-
Download the pretrained YoloV5 model available at this link, and move it into the directory "Utils/yolov5"
-
Check everything is correctly installed by running the "main.py" in the "PLOP" directory
The PLOP algorithm can be run for learning to recognize the following object properties: dirty, open, filled, toggled (for further details about the learning tasks, please see the paper). The learning task can be changed by setting the "TASK" flag in "Configuration.py".
e.g. to run PLOP on the task of learning to recognize the property "dirty", set TASK = TASK_LEARN_DIRTY in "Configuration.py"
When you execute PLOP, a new directory with all logs and results is created in the "Results" folder. For instance, the logs and results are stored in the folder "Results/test_set_X_stepsY", where X is the task name set in "Configuration.py", and Y the number of steps (which can be set in "Configuration.py"). One subdirectory is created for each episode, which consists of a run in a single environment. Each episode subdirectory contains evaluation and log files relative to a single episode.
- The training sets collected online for each property and used for evaluation in the AAAI 2023 paper can be downloaded at this link
If you find this work useful, please consider citing the following papers.
@inproceedings{Lamanna_AAAI_2023,
title={Planning for Learning Object Properties},
author={Lamanna, Leonardo and Serafini, Luciano and Mohamadreza, Faridghasmenia, and Saffiotti, Alessandro and Saetti, Alessandro and Gerevini, Alfonso Emilio and Traverso, Paolo},
booktitle={Proceedings of the 37th AAAI Conference on Artificial Intelligence},
year={2023}
}
@inproceedings{lamannaonline,
title={Online Grounding of Symbolic Planning Domains in Unknown Environments},
author={Lamanna, Leonardo and Serafini, Luciano and Saetti, Alessandro and Gerevini, Alfonso and Traverso, Paolo},
booktitle={19th International Conference on Principles of Knowledge Representation and Reasoning},
year={2022}
}
This project is licensed under the MIT License - see the LICENSE file for details.