Motion and Hardware Intelligence for the Adaptive Morphology of Flying Multibody Robots

Student thesis: Phd


Adaptive morphology is the animals' ability to change body shape to accommodate various dynamic requirements. This ability has inspired the development of robotic platforms with varying morphology, offering novel functionalities and higher adaptability to different environments. Morphing robots have the potential to overcome the challenges faced by existing robots, creating platforms that can operate in complex and unpredictable domains. There already exists proof-of-concept of morphing robots that, through their shape-changing behaviour, have demonstrated multimodal locomotion strategies and higher efficiency in completing duties. Most studies have primarily focused on demonstrating the morphing capability at the hardware level, often leaving motion intelligence as a secondary, seldom addressed problem. In such cases, the shape-changing capabilities are remotely controlled by humans or through basic control strategies where the actuation commands are proportional to measured variables. From the hardware perspective, the success of the shape-changing behaviour relies on the designer's ideas and capabilities, and the final design is often refined through an iterative process that may involve physical prototyping and testing. This design process often leads to a suboptimal design with limited embodied intelligence. In contrast, this manuscript takes a different route. Given a conceptual design, we immediately address the motion intelligence problem, which enables us to assess the motion and control capabilities providing valuable insights for the design of future versions. In this thesis, we discuss the integration of morphing capabilities in two platforms: i) a flying humanoid robot, which represents, to the best of our knowledge, the first application of adaptive morphology to such a nature of multimodal platform; and ii) a small drone with state-of-the-art morphing wing strategies. For both morphing systems, we propose a systematic modelling approach. Then, we introduce a degree of motion intelligence by employing model-based optimisation techniques including in the analysis factors such as energy consumption, aerodynamic efficiency, hardware limitations, and actuation redundancies. For the flying humanoid robot, we propose the implementation of robotic covers that can actively change their shape to control the aerodynamic forces acting on the body, enhancing the robot's flight efficiency. We validate the control law and the morphing capabilities through testing in a custom-made simulation environment, and the results provide evidence of the cover hardware and motion intelligence to morph as desired. For the drone, we present a novel framework that autonomously designs morphing drones with integrated embodied intelligence using evolutionary algorithms. The drone designs are optimised based on agility, manoeuvrability, and energy efficiency criteria. We test the designed morphing drones in simulation scenarios and compare their performances with a commercial fixed-wing drone. The results demonstrate the effectiveness of morphing and the framework's ability to propose functional solutions.
Date of Award31 Dec 2023
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorDaniele Pucci (Supervisor) & Angelo Cangelosi (Supervisor)


  • adaptive morphology
  • modelling
  • aerial robotics
  • hardware optimisation
  • autonomous aerial vehicles
  • humanoid robots
  • morphing robots

Cite this