GridCell-3D

This repository contains a tensorflow implementation of 3D grid cell, which is in the supplemental material of paper "Learning Grid Cells as Vector Representation of Self-Position Coupled with Matrix Representation of Self-Motion".

Reference

@article{DG,
    author = {Gao, Ruiqi  and Xie, Jianwen and and Zhu, Song-Chun and Wu, Ying Nian},
    title = {Learning Grid Cells as Vector Representation of Self-Position Coupled with Matrix Representation of Self-Motion},
    journal={Seventh International Conference on Learning Representations (ICLR)},
    year = {2019}
}

Requirements

• Python 2.7
• Tensorflow r1.0+

How to run

(0) Learning 3D units

Training with single block setting

$ python ./code_single_group/main.py --single_block True --num_group 1 --block_size 8

Visualization

$ python ./code_single_group/visualize_3d_grid_cells.py

(1) Path integral

Training with Gaussian kernel

$ python ./code/main.py --mode 0 --GandE 1

Testing path integral

$ python ./code/main.py --mode 2 --GandE 1 --ckpt model.ckpt-7999

(2) Path planning

Training with exponential kernel

$ python ./code/main.py --mode 0 --GandE 0

Testing path planning

$ python ./code/path_planning.py    

Testing path planning with obstacle

$ python ./code/path_planning_obstacle.py

Q & A

For any questions, please contact Jianwen Xie ([email protected]) and Ruiqi Gao ([email protected])