UGV-Surveillance-with-Coordinated-Behavior-Using-Behavior-Trees

A multi‑robot Unmanned Ground Vehicle (UGV) system using behavior trees for coordinated surveillance, obstacle avoidance, and target(human) detection.

Red Cube detection using OpenCV and Human Detection using YOLOv5

Obstacle Detection and Update

Once a red cube is detected, it updates as an obstacle for every robot in their costmap as a pink circle.

Navigation Towards the human

If any person is detected, the robots communicate and all the robots navigate towards the person.

Working Video: Videos

---

Steps to run the package

1. Build the Package

   colcon build
   

2. Source the setup file

   source install/setup.bash
   

3. Initial launch

   launches the world with robots and also launches NAV2

   ros2 launch inception_p0 gazebo_multi_nav2.launch.py
   

4. Open a new terminal

   launches the red cube detection, yolo human detection and updates the detected red cubes as obstacles for all robots

   ros2 launch inception_p0 global.launch.py
   

5. New terminal

   launch the behavior tree

   ros2 launch inception_p0 autonomy.launch.py
   

Description

• On launching all the launch files, the robots will start moving around the map to the given locations in locations.yaml file.
• The robots will add the red cubes as a obstacles in the costmap which will update for both of the robots. The pink circles depict the red cubes added as obstacles.
• While navigating around the map if the robot detects a human using the YOLO model, the robots will stop navigating and go towards the person.

Behavior Tree Overview

The behavior tree is defined as following: - Parallel Node runs two tasks simultaneously 1. 'CheckForPerson' detects if a person is present 2. 'navigate_sequence': navigates sequentially to predefined locations in the .yaml file. - 'GoToPose (navigate_to_person): If a person is detected, the robot will navigate to the person's location.

Customization

• Change the spawn location of the robot in the gazebo_multi_nav2.launch.py.
• We can change the radius of the hazardous zone in the red_cube_detect.py.