epoxx-arch / ood_project

Out of distribution detection using simulation-informed deep Gaussian process state-space models

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Introduction

Software infrastructure for out-of-distribution detection with a real quadrupedal robot based on Gaussian process state-space models. This repository implements the experiments of the paper:

Out of Distribution Detection via Domain-Informed Gaussian Process State Space Models Alonso Marco, Elias Morley, Claire J. Tomlin 2023 IEEE 62nd Conference on Decision and Control (CDC)

daisy

It features as prediction module (C++) for fast rolling-out of state trajectories. Implemented using Eigen and pybind11

It needs the auxiliar repository LQRker.

Installation notes

conda create -n ood python=3.7

pip install tensorflow
pip install tensorflow_probability
pip install numpy scipy matplotlib==3.2.2
pip install hydra-core --upgrade
pip install control
pip install -e .
cd <path/to/LQRker>
pip install -e .

About

Out of distribution detection using simulation-informed deep Gaussian process state-space models


Languages

Language:Python 97.4%Language:C++ 2.2%Language:CMake 0.3%