UC San Diego

In the last decade, roboticists have had significant success building centimeter-scale flapping wing micro aerial vehicles (FWMAVs) inspired by the flight of insects. Evidence suggests that insects store and release energy in the thoracic exoskeleton to improve energy efficiency by flapping at resonance. Insect-inspired micro flying robots have also leveraged resonance to improve efficiency, but they have discovered that operating at the resonant frequency leads to issues with flight control. This research seeks to investigate the roles that elasticity, aerodynamics, and muscle dynamics play in the emergent dynamics of flapping flight by studying elastic flapping spring-wing systems using dynamically-scaled robophysical models of spring-wings. Studying the dynamics of a robot with comparable features enables the validation of models from biology that are otherwise difficult to test in living insects, the generation of new hypotheses, and the development of novel FWMAV designs. In Chapter 1, the spring-wing system is characterized as a nonlinear spring-mass-damper model. A robophysical model validates that such systems gain energetic benefits from operating at resonance, but reveals that the benefit scales with an underappreciated dimensionless ratio of inertial to aerodynamic forces, the Weis-Fogh number. We show through dimensional analysis that any real system, living or robotic, must balance the mechanical advantage gained fromoperating at resonance with diminishing returns in efficiency. Chapter 2 further explores the impact of the Weis-Fogh number on flapping dynamics, showing that responsiveness to control inputs is reduced and resistance to environmental perturbations is increased as the dimensionless ratio increases. Together with calculations of Weis-Fogh number in insects, these studies illustrate tradeoffs that drive evolution of resonant flight in nature and guide development of future FWMAVs with elastic energy exchange. In the second half of the thesis, muscle dynamics are introduced in the form of a simplified model of self-excited asynchronous insect muscle. In Chapter 3, a linear feedback model adapted from experiments on insect flight muscle is developed and integrated with the spring-wing model, producing a system that generates steady flapping via limit-cycle oscillations despite the absence of periodic control inputs. The model is explored analytically, in simulation, and via implementation on the robotic spring-wing. Novel dynamic characteristics that enable adaptation to damage and passive response to wing collisions are described. Chapter 4 leverages the asynchronous feedback model as part of an interdisciplinary study of the evolution of asynchronous muscle. Phylogenetic analysis, direct measurement of insect muscle dynamics, and experiments on the robophysical system show that evolutionary transitions between periodically forced and self-excited insect muscle were likely made possible by a ”bridge” in the dynamic parameter space that could be traversed under specific conditions. The asynchronous spring-wing model provides new insight into the flight and evolution of some of the most agile insects in nature, and presents a novel adaptive control scheme for future FWMAVs.