Grid Cell-Inspired Fragmentation and Recall for Efficient Map Building

Official implementation for Grid Cell-Inspired Fragmentation and Recall for Efficient Map Building (FARMap) in TMLR 2024 (Featured Certification; 3.5% among accepted papers). We propose a new framework for mapping based on Fragmentation-And-Recall, or FARMap, that exploits grid cell-like map fragmentation via surprisal combined with a long-term memory to perform efficient online map building. Our experiments show that FARMap reduces wall-clock time and the number of steps (actions) taken to map large spaces, and requires a smaller online memory size relative to baselines.

Run Experiments

Setup

pip install -r requirements.txt

(Optional) Environment Generation

python3 environment_generation/generator.py

Exploration

python3 run_exp.py -wandb -env $ENV_ID

Citation

If you use this code for your research, please cite our paper.

@article{hwang2024grid,
    author = {Hwang, Jaedong and Hong, Zhang-Wei and Chen, Eric and Boopathy, Akhilan and Agrawal, Pulkit and Fiete, Ila},
    title = {Grid Cell-Inspired Fragmentation and Recall for Efficient Map Building},
    journal={TMLR},
    year = {2024},
    url={https://openreview.net/forum?id=cT8oOJ6Q6F},
    note={Featured Certification}
}