Perception-aware Drone Path Planning for Monitoring and Assessment in Disaster Response

Drones, especially multi-rotors, have become a crucial tool for providing situational awareness to emergency service providers in their effort to tackle unforeseen and critical events like forest fires, floods. In such events, the drone is launched from a location that is not far away from the disaster region, to fly around the location to survey and monitor the region and collect relevant information and aerial data for further assessment. The main challenge is that the drone is required to have a flight path that starts from its starting point, cruise around the region, and return to its initial position. This becomes a highly complex task to achieve effectively by a human pilot controlling the drone considering the situation risks. A perception-aware drone path planning can enhance the monitoring effectiveness and increase the survey and collected data accuracy. Using onboard visual capabilities, the drone can autonomously detect the region of interest, fly around and follow the boundary, while relaying real-time information to the disaster response team for immediate assessment without physically and mentally overloading a human pilot to control the drone.;;This project aims to address the automation of the path planning problem for surveying and monitoring with the ability to avoid obstacles along the established flight path and handle weather conditions that may impact the flight and the accuracy of the collected data. In this context, the research considers:;;Dynamic and real-time path planning using onboard visual capabilities considering monitoring and survey parameters including flight speed, flight altitude, monitoring standoff distance and direction etc.;Image processing segmentation and machine learning techniques to identify fire regions and boundaries and construct real-time geometric models for use to adapt the pre-established flight path;Obstacle avoidance and wind condition such as gusts and wind direction that may deter the drone from its designated path.