Baselines forked from CoRL 2022 Paper: Visuo-Tactile Transformers for Manipulation
In this project, we present a novel model for fusing Visuo-Tactile data for robot manipulation task. We propose:
Convolutional Layer for visual data stream.
LSTM for tactile data stream
Attention layer for fusing the both streams.
Baselines for fusing Visuo-Tactile data:
Visuo-Tactile Transformer (VTT)
Product-of-Experts (POE)
Concatenation (Concat)
Requirements:
torch==1.9.0
tqdm==4.48.2
pybullet==3.1.8
gym==0.17.2
matplotlib==3.4.3
numpy==1.21.2
pandas==1.1.2
Minitouch Installation:
cd Minitouch
pip install -e .
Available Tasks:
Pushing-v0
Opening-v0
Picking-v0
Inserting-v0
Example of Running Code:
python train.py --encoder="VTT" --seed=0 --task_name="Pushing-v0"
Read Results with read_pickle.py
For detailed information check Report.pdf
Orignal Codes:
- Yizhou Chen, Andrea Sipos, Mark Van der Merwe, Nima Fazeli
- https://github.com/yich7045/Visuo-Tactile-Transformers-for-Manipulation