Enhancing autonomy of aerial systems via integration of visual sensors into their avionics suite

Abstract

Autonomous aerial systems have started to gain much traction in the military intelligence, surveillance and reconnaissance domain. The remotely piloted systems, such as the Predator, are already successful unmanned systems; the next step forward is to use autonomous systems to overcome high manning requirements. These systems are scalable and serve as excellent force multipliers, but there are other technological issues to overcome to qualify an autonomous aerial system, such as navigation and collision avoidance. This thesis explores autonomous system capabilities using quadrotors in the context of the Singapore Armed Forces. It first applies a systems engineering approach to analyze stakeholders' needs, then translates the needs to functional requirements, and concludes with the development of a possible system architecture for an autonomous quadrotor system. The author then conducted indoor flight experiments to validate the capabilities of waypoint navigations and collision avoidance. The results were highly encouraging and qualified the aerial platform for subsequent visual sensor integration.