Skip to content

PratikKunapuli/AerialManipulation

Repository files navigation

AerialManipulation

IsaacSim IsaacLab Python

Repository for exploring Reinforcement Learning (RL) approaches to control of a 2 degree of freedom (DoF) aerial manipulator, written for IsaacLab

Installation

Pre-requisites

Conda environment

conda create -n isaaclab python=3.10
conda activate isaaclab

IsaacSim

Instructions for installing IsaacSim are reproduced here for clarity - but you should follow the official instructions for any troubleshooting.

Python environment installation
pip install isaacsim==4.1.0.0 --extra-index-url https://pypi.nvidia.com

(Optional)

pip install isaacsim-extscache-physics==4.1.0.0 isaacsim-extscache-kit==4.1.0.0 isaacsim-extscache-kit-sdk==4.1.0.0 --extra-index-url https://pypi.nvidia.com

IsaacLab

Installations for installing IsaacLab are reproduced here for clarity - but you should follow the official instructions for any troubleshooting.

Installing IsaacLab

Clone the repo locally

git clone git@github.com:isaac-sim/IsaacLab.git

Install dependencies via apt

sudo apt install cmake build-essential

Install the library (this should find the previously created conda env isaaclab since it is the default name, but if you changed the name for the conda environment you can specify the env name in this command)

cd IsaacLab
./isaaclab.sh --install

AerialManipulation Installation

Clone the repo

git clone git@github.com:PratikKunapuli/AerialManipulation.git

Install locally

cd AerialManipulation
pip install -e .

Install MySQL only for creating a local database with GC tuniung

sudo apt-get install mysql-server

Demos

Two demonstration files are available to investigate the model/physics.

  1. demo_sim.py: A script used to launch an IsaacSim window where the dynamics are simulated directly
  2. demo_env.py: A script used to launch an IsaacSim window from the gymnasium api.

Environments

Training

Tools

Converting URDFs to USD
python ./IsaacLab/source/standalone/tools/convert_urdf.py ./AerialManipulation/models/aerial_manipulator_2dof.urdf ./AerialManipulation/models/aerial_manipulator_2dof.usd --merge-joints --make-instanceable

About

Repository for exploring Reinforcement Learning (RL) approaches to control of a 2 degree of freedom (DoF) aerial manipulator, written for IsaacLab

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors