Leveraging Aerial Transformation for Enhanced Air–Ground Robotic Mobility

Ground-aerial robots can extend endurance, versatility, and robustness by combining wheeled motion with flight, yet many flying-rolling robot designs add actuators that increase weight and reduce efficiency. Morphobots mitigate this by using multi-purpose actuators and body shape change to switch modes on the ground, but unpredictable vehicle-ground interactions can be an obstacle to robust operation. This dissertation develops the Aerially Transforming Morphobot (ATMO), a quadcopter that reconfigures in flight to land on wheels, enabling reliable air-ground transitions, mode switching without the hindrances of ground-morphing, and improved agility. We present ATMO’s design and performance characterization, analyze its dynamics–revealing transformation-induced couplings incompatible with standard quadcopter control–and introduce a model-predictive control framework that stabilizes ATMO through aerial transformation to execute dynamic transitions. We then compare this approach with a learning-based controller that uses deep reinforcement learning for end-to-end morpho-transition, validating both experimentally. Finally, we revisit ATMO’s design using aerodynamic principles to expand morphing flight through wake vectoring, showing that passive structures in the rotor wake substantially increase available thrust authority. Overall, we demonstrate that aerial shape change improves agility and reliability, highlighting a new direction for research in ground-aerial robotics.