Quadcopter Simulation Model for Research of Monitoring Tasks

摘要

The article describe a simulation of the drone designed to monitor a large area. Three scenarios are considered: detection of the fire source in the forest, detection of intruders in the forbidden territory and detection 0 their cars. Open source software is used for the simulation: robotic simulator Gazebo, framework for robotic applications Robot Operating System, communication protocol for small unmanned vehicles MAVLink and autopilot system PX4. Iris+ quadcopter is chosen as a prototype for the simulation, its mathematical model is presented as well as the vision model based on the PX4FLOW monocular camera and optical flow. Algorithms for detecting objects of interest are described. As a result, the successful tests of simulation are presented, in which the classification and localization accuracy is tested.