Final Project PDM Group 6 "Obstacle driven RRT algorithm for quad-rotor motion planning"

Group members:

• Andrei-Carlo Papuc, 4772385
• Benjamin Bogenberger, 5845033
• Crina Mihalache, 4827333
• Simon Gebraad, 4840232

About this repository

This repository encapsulates a quad-rotor motion planning stack with various implementations for offline planning ((biased) RRT, (biased) RRT*, (biased) informed RRT*, PRM) and online MPC control. This stack is integrated in a physics simulation with multiple environments. On top of that evaluation scripts and results are provided.

This repository was developed using Python 3.8. Its dependencies are:

• NumPy
• Matplotlib
• scikit-learn
• CVXPY
• gym-pybullet-drones (v1.0.0)
• tqdm
• ipykernel

The repository is structured as follows.

• src/ contains the main code
• main_simulation.py executes the full planning and control stack in a physics simulation
• main_rrt.py, main_prm.py and main_rooms.py show their respective planning part
• the Jupyter notebooks contain code for evaluation of the planning algorithms
• saved_results, saved_results_hole, saved_results_simulation and plots contain main evaluation data
• linearize_quadrotor_model.m symbolically linearizes a quad-rotor model

Setting up python environment

Set up the conda environment using

conda env create -f environment.yml
conda activate pdm_drone
pip3 install --upgrade pip
cd ..
git clone https://github.com/utiasDSL/gym-pybullet-drones.git
cd gym-pybullet-drones/
pip3 install -e .

Run applications

• Full simulation (Planning & MPC):

python3 main_simulation.py (--room 0...3) (--planner RRT,RRT_s,iRRT_s,PRM)

• Planning applications alone ((B)RRT, (B)RRT*, (B)iRRT*, PRM): main_rrt.py and main_prm.py