While much has been written about quadrotor MAV (micro aerial vehicle) theory and operation, there is relatively little that communicates the practical steps necessary to build, program, and control these complex electromechanical systems. Research abounds that deals with the minutia of control and modelling, but the question of minimal sufficiency is left unanswered and unproven on real hardware. This thesis demonstrates that PID control and operational models that are simple in form but relevant to real time operations are all that is necessary to achieve stable autonomous flight. Furthermore, it presents a clear and practical approach to the development of real flying robots.

This work describes the construction of three functional MAVs that cover a range of size, complexity, and functionality. Details of airframe structure, component selection, firmware and software development, and tuning and testing are all included and they provide a reproducible framework for further research.

The culmination of this effort takes the form of a real physical (not simulation), fully autonomous quadrotor MAV of non-trivial mass (greater than 2kg), payload capacity (theoretically greater than 2kg), and computing power (running Linux on a processor capable of up to 1400 Dhrystone MIPS). Its successful operation is presented and serves to demonstrate the efficacy of the proposed straightforward, minimalist approach to design and development.