VAYUBHOOMI UAV UGV PIL PROJECT 2022

Networked autonomous UAV - UGV system for reconnaissance and waste management over open spaces.

Approach:

Our approach relies on three main sub-systems :

• The Unmanned Aerial Vehicle(UAV) : Patrols preset area using :
  • Proportional Integral Derivative(PID) controller
  • records video and pose
  • sends to base station.
• The base station:
  • Per frame video feed analysis by Neural net(NN)
  • Neural Networks Detects trash
  • Filters trash positive frames
  • K-means cluster location stored in CSV file.
• The Unmanned Ground Vehicle(UGV):
  • Collects cluster location and moves to them;
  • Scans cluster vicinity for accurate trash detection.

Frameworks and Languages used:

• ROS
• PYTHON
• XML

Note:

• Files wrt to simulation is located under lidar package.
• Files wrt to trash detection and clustering is present in vision package.

Gazebo worlds used for simulation:

Procedure to launch the simulation:

• Make sure you have installed models necessary for trash.world from (https://github.com/osrf/gazebo_models)

• Make sure the .world files are present in worlds folder of your ROS package and all the models must be present in the .gazebo/models folder for the launch file to work.

• Update the following files :

  • droneobsavoid.py

  • botmove.py

  • dronespiral.py

  • UAVpathRecorderNode.py

• Command to launch the simulation :

  roslaunch lidar sample.launch

  Change lidar to the name of your ROS package