The current crop of outdoor-focused quadrotors struggle to explore tight, GPS denied, and vision impaired environments while managing self-induced turbulence and keeping the environment and UAV itself safe from collisions. Mainstream designs typically arrange their propellers in or near the same plane, resulting in an under-actuated system that must roll and pitch in order to move laterally.
This dissertation describes the design, analysis, and construction of a soft multirotor airframe with the capability of in-plane maneuverability and decoupled 6DOF control which allow for low profile sensor payload integration without the need for a gimbal as well as predictable and safe flight in confined spaces such as tunnels and collapsed buildings. All aspects of the design are described, beginning with the electronics package itself.
The author observes the disappointingly low use of small embedded Linux platforms in robotics education. As an alternative to the ubiquity of microcontroller-based development boards such as Arduino, this work presents the use of the Robotics Cape and BeagleBone Blue along with their associated software and hardware ecosystem in both a prototyping and education environment. This ecosystem was initially designed and produced to facilitate the aforementioned multirotor's construction, yet continues to be used as part of the UCSD MAE curriculum for the benefit of others.
Following the design of the multirotor's electronics package, this dissertation presents a method for optimizing rotor angles as a function of the frame parameters and the desired performance characteristics. It compares the performance of an optimized configuration with an existing commercial hexacopter, and quantifies the improved control authority of the optimized design.
The multirotor concept is then implemented as a monocoque airframe, designed through a presented technique for rapidly iterating the airframe shell thickness based on modal analysis using finite shell elements. Finally, the real-world performance of the platform is evaluated by examining several close-quarters flight scenarios using CFD, through analytical performance characterization of the supporting flight controller, and through physical measurement of the constructed multirotor's control authority.
