OPTIMIZING THE STRUCTURED FLIGHT TRAJECTORY FOR DRONES TO INSPECT HDB BUILDING FACADES

Abstract

The optimization of a drone’s flight trajectory has already seen great commercial success in terms of horizontal flight planning using latitude and longitude values. However, most are still lacking in vertical flight planning tools as there is a need to account for altitude values. Also, several considerations must be given when planning for an autonomous vertical flight trajectory. This includes the pattern of flight, drone’s specifications, required camera and image parameters, as well as the need to convert between local Cartesian and absolute world coordinates based on the geographic coordinate system.

The planning of a structured and autonomous flight trajectory for a drone to carry out façade inspection contains several modules written in Python language. Each module contains a specific function and can produce file outputs in .csv format. This allows interoperability with other programs that will assist with the overall façade inspection process. These other programs include 3D remodeling of the building, resolution of the Travelling Salesman Problem, flight visualization and interfacing with a self-developed mobile application.

Based on the test cases of this study, the results are optimistic in generating a three-dimensional flight trajectory that allows full maneuverability. The flight trajectory can be aligned with the actual location and orientation of the inspected building. In addition, the flight program is capable of inspecting building façades by calculating the optimal distance of the drone to the façade with considerations given to the required camera and image parameters.

It is concluded that the conceptual framework for an autonomous flight trajectory program is achieved, albeit with several enhancements still to be made. Further developments can be explored to improve the adaptability of the flight program. This includes customization of the camera’s gimbal angle along the flight, recalling and resumption of flight based on the last waypoint visited, and finally an optimization of the program in terms of space and time complexity. It is also possible to investigate how the structured flight pattern can be rerouted around no-fly zones based on user input in the program.