PRIMAL Installation directions

Prerequisites

Git
Anaconda
Boost
If using GPU:
- cuDNN 6
- CUDA 8
- GCC 4.8

Installation steps

Run this command: conda env create -f environment.yml (or, if using a GPU conda env create -f environment_gpu.yml) This will create a new conda environment called primal that includes all the dependencies
Activate the environment with this command: conda activate primal
Run cd od_mstar3 and then python3 setup.py build_ext --inplace, this builds the cpp_mstar library. (Note: if python3 in the previous command does not work, try using just python)
Run cd .. to get back in the main directory

Installing pre-trained model

Make sure you're in the root directory
Create a directory called model_primal if it does not exist
Download this file: https://drive.google.com/file/d/1AtAeUwLF1Rn_X3b2FHkHi4fI5vveUHF6/view (it will take some time)
Extract the contents of that file to the model_primal directory

You're all done!

Notes:

Training is done by the DRLMAPF_A3C_RNN.ipynb notebook. I do reccommend using the pre-trained model instead of retraining. Do not train unless you know what you're doing.
To use the map creator, just run python mapgenerator.py
Hotkeys for map generator:
- o: obstacle mode
- a: agent mode
- g: goal mode, click an agent then click a free tile to place its goal
- c: clear agents
- r: reset
- up/down arrows: change size
- p: pause inference