Planning and decision making final project

Tatsuki Fujioka, 5849837; David Schep, 5643384; Dielof van Loon, 5346894

Install

Install https://github.com/utiasDSL/gym-pybullet-drones using their installation instructions:

git clone https://github.com/utiasDSL/gym-pybullet-drones.git
cd gym-pybullet-drones/

conda create -n drones python=3.10
conda activate drones

pip3 install --upgrade pip
pip3 install -e . # if needed, `sudo apt install build-essential` to install `gcc` and build `pybullet`

Run

Always activate the conda environment before running code: conda activate drones.

To run the main script that generates the data used in the project:

python3 compare_algos.py

This runs 200 simulations. 100 for RRT and 100 for RRT-u. This takes approximately 20 minutes, depending on processing power.

The compare_algo.py script has a few command line parameters:

  --seed_start SEED_START
                        The start seed for the tests. (default=1240918)
  --num_seeds NUM_SEEDS
                        The number of seeds to test. (default=100)
  --gui GUI             Whether to use GUI or not.
  --num_restarts NUM_RESTARTS
                        The number of restarts each algorithm is allowed. (default=0)
  --num_obstacles NUM_OBSTACLES
                        The number of obstacles. (default=30)
  --sample_box SAMPLE_BOX
                        The sample box for the algorithms. (default=((-2, -2, 0), (6, 6, 4)))

For example, to run a single seed, with gui:

python3 compare_algos.py --gui True --num_seeds 1 --seed_start 1234

This runs both RRT and RRT-u on the same seed, once.

RRT and RRT-u visualization scripts

Both rrt and rrt-u have a visualization script. They can be run using the following commands:

python3 vis_RRT.py
python3 vis_RRTu.py

These scripts only run the path planning algorithm and do not fly the drone along the path. They are used to collect images and videos for the report and presentation.

File structure

• Obstacles.py: Contains the Obstacles class, generates the random obstacles.
• util.py: Contains utility functions.
• create_graphs.ipynb: A jupyter notebook to generate a graph for the report.
• all_test_PID.py: Contains the simulator. This simulator function takes callback functions that calculate the path using RRT or RRT-u.
• RRT.py: Contains the RRT class, uses the RRT algorithm to generate a path.
• RRTu.py: Contains the RRT-u class, uses the RRT-u algorithm to generate a path.
• vis_RRT.py: Used te create visualizations of RRT
• vis_RRTu.py: Used te create visualizations of RRT-u