Autonomous Drone Navigation in an Urban Environment

Teaser.mp4

Hadar Hai, Technion - Israel Institute of Technology

The objective of this project is to develop a codebase that enables autonomous navigation for drones within urban settings using a trajectory planning algorithm. The autonomous drone will initiate its flight from a designated start position and maintain a constant height while navigating the most efficient route towards a specified goal position. Upon reaching the goal, the drone will enter a hovering state, signifying its arrival at the destination.

The drone's LIDAR sensor will be utilized to detect obstacles obstructing its flight path, thereby preventing a direct route to the goal position. Upon identifying such obstacles, the drone is required to implement an evasion strategy, circumventing the obstruction until a clear pathway to the goal is achievable.

Table of Contents

• Requirements
• How to Use
• Examples
• Sources
• Acknowledgements

Requirements

The code is intended to be used on Windows 10 machines, and on a compact version of AirSim.
Therefore, if the normal AirSim installation does not work, please contact us for the slim version.

Install the Python environment.

1. Create an environment.

python -m venv venv 
venv\Scripts\activate.bat
pip install wheel

2. Install the AirSim Python API dependencies.

pip install numpy msgpack-rpc-python matplotlib pandas shapely pyqt5 

3. Install AirSim Python API.

pip install airsim

1. Install conda.
2. Create a conda environment: conda create --name autDrone python=3.8
3. Activate the environment: conda activate autDrone
4. Install the AirSim API package: pip install numpy msgpack-rpc-python matplotlib pandas airsim

How to Use

1. Make sure that the AirSim application is up and running.
2. Run main.py using your desired path, e.g.:

python main.py --start_x -700 --start_y -1100 --end_x -1216 --end_y -372 --height -50

Examples

We provide in the Examples directory several navigation examples in video format, using different heights and different paths used, for the following navigation paths:

Sources

• Kamon, Ishay, Elon Rimon, and Ehud Rivlin. "Tangentbug: A range-sensor-based navigation algorithm." The International Journal of Robotics Research 17, no. 9 (1998): 934-953.

Acknowledgements

We thank Hila Manor for contributing the needed resources for running the project.
The project was created as a part of course CS236927 of Computer Science faculty, Technion.