Aero Scan

[jayhaych02]

https://github.com/jayhaych02/AeroScan - Slideshow for Presentation & Proof of Concept Video are included

Project Abstract

AeroScan is a simulated autonomous security patrol drone designed to navigate predefined areas, detect intruders, and avoid obstacles using sensors like 2D LiDAR and an Intel RealSense depth camera. The simulation will be built in Gazebo, integrated with ROS2 Humble for sensor processing, navigation, and obstacle avoidance. PX4 SITL will be utilized for simulating the drone's flight dynamics, allowing for realistic command execution, such as takeoff and landing. This project focuses on developing a fully simulated environment where the drone patrols a security area, detects intruders, and sends real-time sensor data for analysis.

Conceptual Design

• ROS2, PX4 SITL, and Gazebo:
• ROS2 Humble: The core of the software architecture, responsible for communication between nodes, sensor data processing, and decision-making.
• PX4 SITL: Provides a realistic simulation of flight dynamics, allowing the drone to execute commands like commander takeoff directly in the PXH shell.
• Gazebo: The drone and environment will be simulated entirely in Gazebo, where the drone will patrol, navigate waypoints, and detect obstacles/intruders. Sensors in Simulation:
  1. 2D LiDAR (RPLiDAR A1):
• Purpose: Detects nearby obstacles and assists in obstacle avoidance while patrolling. Provides real-time distance measurement for 360-degree surroundings.
• Integration: Simulated using a Gazebo plugin for 2D LiDAR sensors, connected to ROS2 for data processing.

2. Intel RealSense Depth Camera:

• Purpose: Provides 3D depth perception and color images to detect obstacles and intruders. Simulates the drone’s front-facing view to monitor the area ahead.

• Integration: Simulated using the realsense-ros package, integrated with Gazebo for real-time data processing in ROS2.

Simulation Environment:

• Intruder Simulation: Add Gazebo “actors” that represent intruders moving in a predefined patrol area (e.g., a warehouse or parking lot).

• Obstacle Avoidance: Implement ROS2 nodes for basic obstacle detection using the 2D LiDAR and RealSense depth camera.

• Patrol Area: The environment will feature barriers or walls that define a secure perimeter. The drone will autonomously navigate between waypoints in this area.

Key Features:

• Autonomous Navigation: The drone will navigate between predefined waypoints using ROS2, with flight dynamics handled by PX4 SITL, and avoid obstacles detected by LiDAR and the depth camera.

• Intruder Detection: Use the RealSense camera or LiDAR to detect intruders in the environment and trigger an alert system when an intruder is found.

Proof of Concept https://github.com/jayhaych02/AeroScan Proof of concept is verifying that the PX4 SITL simulated drone taking off via the px4 shell “px4>” inside a ROS2 Humble Linux virtual machine works.

Background This project builds on current trends in autonomous drones used for security patrols and surveillance. Similar commercial solutions are either closed-source or prohibitively expensive due to the use of complex hardware and software. AeroScan focuses on using affordable components like 2D LiDAR and simulated environments to reduce costs and make security patrol systems accessible.

Required Resources

Software Resources:

• ROS2 Humble: The core framework for communication between sensor nodes, navigation, and obstacle detection.
• PX4 SITL: For simulating flight dynamics and control.
• Gazebo: Used for creating the simulation environment.
• Intel RealSense ROS2 Package: For depth camera integration.
• RPLiDAR ROS2 Package: For simulating 2D LiDAR.

[chiikugo]

Never worked with hardware, but from your demo, I think this would be a really interesting project to work on. The simulation you launched in class and even to working on the drone is a new experience for many of us.

Would love to work on the automation of this project!