Developing real-time uncertainty quantification for semantic segmentation tasks. The goal is to have a downstream robot, conditioned on well calibrated UQ statistics interact with human operators on joint assembly tasks.
maintainer: slwanna@utexas.edu
This project assumes you have CUDA support 11.6+ and use anaconda for python environment management.
- Clone the github repository:
$ git clone https://github.com/UTNuclearRoboticsPublic/rt_uq_semantic_segmentation- Recreate the python environment
$ conda env create -f environment.yml- Install StochMan dependency in project directory
$ git clone https://github.com/IlMioFrizzantinoAmabile/stochman
$ cd stochman
$ python setup.py install