Towards reverse engineering the neuromechanical actuation and control of biological flapping flight
Restricted (Penn State Only)
- Author:
- Agrawal, Suyash
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 27, 2024
- Committee Members:
- Jean-Michel Mongeau, Major Field Member
Puneet Singla, Outside Unit & Field Member
Chris Rahn, Co-Chair & Dissertation Advisor
Bo Cheng, Co-Chair & Dissertation Advisor
Robert Kunz, Professor in Charge/Director of Graduate Studies - Keywords:
- Flapping wing flight
Wing neuromechanics
Flapping wing robot
Flapping wing simulation
Flapping wing energy efficiency
Hummingbird flight
Insect flight
Robotic flight
Flapping wing actuation
Flapping wing control - Abstract:
- Natural fliers use their flapping wings to produce remarkable flight maneuvers, achieving a level of aerial maneuverability largely unattainable by robotic fliers. These wings are moved by their wing neuromechanical systems, which consists of several biomechanical sub-systems such as musculoskeletal actuation system, central and peripheral motor circuits, and proprioceptors. While extensive research has been conducted on each of these components, a comprehensive understanding of this system, specifically how this system actuates and controls the wing while maintaining high energy efficiency, is still lacking. Such understanding is also critical for informed development of flapping wing robots with performance comparable to those of natural fliers. Therefore, this dissertation aims to advance our understanding of wing neuromechanical systems in natural fliers, particularly focusing on the actuation and control, as well as on trade-off between energy efficiency and active control. This dissertation initially focuses on studying the hummingbird musculoskeletal (wing actuation) system. A functional model of this system is developed by synthesizing existing empirical data and literature. The model is used to predict instantaneous, three-dimensional torque produced by different muscles, and the primary muscle contractile behavior, including stress, strain, elasticity, and work. The results suggested that: i) the primary muscles function as diverse effectors, as they do not simply power the stroke, but also actively deviate and pitch the wing with comparable actuation torque, ii) the secondary muscles produce controlled-tightening effects by acting against primary muscles in deviation and pitching, iii) hummingbirds may be using significant elastic energy, as indicated by their power muscle work loops, and that they may be facing trade-offs between muscle energy efficiency and active control. The comparison of the predicted work loops with those of insects and other birds motivated me to develop a generalized functional model and conduct a literature survey of their wing neuromechanical systems. Next, similar to the hummingbird wing musculoskeletal model, parsimonious functional models of the wing motor system with either synchronous or asynchronous power muscles were developed. The model was then non-dimensionalized and simulated to examine model characteristics as functions of Weis-Fogh number and dimensionless flapping frequency. For synchronous power muscles, the model predicts that energy efficiency trades off with frequency control rather than amplitude control at high Weis-Fogh number; however, no such trade-off was found for models with asynchronous power muscles. The work loops alone are insufficient to fully capture wing motor characteristics, therefore fail to directly reflect the trade-offs. Finally, the simulation results were used to predict motor characteristics for hawkmoth, hummingbird, and bumblebee. Next, a thorough literature survey of wing neuromechanical systems in locusts, hawkmoths, flies, and hummingbirds is conducted to develop their functional architectures, i.e., Dual Neural-Mechanical Oscillator (O_Neuro-O_Mech) and Neurally-modulated Mechanical Oscillator (O_Mech (Neuro)), that correspond to fliers with synchronous and asynchronous power muscles, respectively. These architectures elucidate the control mechanisms used by the natural fliers and are used to hypothesize the key traits of wing neuromechanical system that contribute to maneuverability. Then, it is hypothesized that O_Neuro-O_Mech and O_Mech (Neuro) respectively correspond to two models of maneuverability, i.e., hummingbird model and fly model, which are characterized by high and low power modulation, strong and weak force vectoring, and phasic and tonic muscle force production, respectively. Finally, the developed hummingbird wing-actuation model and the knowledge of neuromuscular control derived via literature survey were combined to develop a simulation model of hummingbird (called SimHummer) that incorporates wing actuation, control, and aerodynamics, body aerodynamics, and head vibration control. SimHummer successfully generated hovering and translational maneuvers. A key feature of SimHummer is that wing kinematics or dynamics are not prescribed, rather they emerge during simulation. The results elucidated the importance of the following in flight stability and control: regulation of wing kinematics, vibration control via neck muscles and tail, role of amplitude and precise timing of muscle active force in flight control, and role of pitch angle feedback in flight control. Lastly, the dissertation draws conclusions regarding the traits of wing neuromechanical systems in natural fliers that should be replicated in the next generation of bioinspired flapping wing robots for achieving animal-level aerial maneuverability.