Model Card Bench-NPIN

This hub contains the pretrained models for Bench-NPIN: Benchmarking Non-prehensile Interactive Navigation.

Checkout our project website for more details.

Model Description

Each of the four folders contains the pretrained baselines for the corresponding non-prehensile interactive navigation task.

Uses

Installing BenchNPIN

1. Clone the project

git clone https://github.com/IvanIZ/BenchNPIN.git

2. Clone the submodules (spfa and GLNS)

cd BenchNPIN
git submodule update --init --recursive

3. Install dependencies.

pip install -r requirements.txt

4. Install Gym environment

pip install -e .

5. Install shortest path module

cd deps/spfa
pip install -e .

Run Pretrained Models

1. Download a pretrained model and place it at the root of the BenchNPIN repository (e.g. ship-ice task)

2. Edit the model path in the scripts/ship_ice_baseline.py file

3. Run baseline evaluation

python scripts/ship_ice_baselines.py