Bioinspired Agile Robotic Jumping and Balancing

Agile locomotion in legged robots requires the coordinated generation, storage, and regulation of energy under highly nonlinear and hybrid dynamics. Inspired by biological systems that exploit compliant structures and aerodynamic appendages, this thesis presents the design, modelling, and control of a bioinspired monopedal robotic platform capable of agile jumping and balancing through the integration of two key mechanical innovations: a Continuously Variable Transmission–enhanced Series Elastic Actuator (C-SEA) for vertical propulsion, and an AeroTail drag-based balancing mechanism for attitude regulation in flight. The C-SEA leg incorporates a non-circular gear-based continuously variable transmission within a series elastic architecture to modulate effective mechanical advantage throughout the jump cycle. This enables improved energy transfer, controllable stiffness characteristics, and enhanced jumping performance compared to conventional SEA configurations. The AeroTail introduces a bioinspired drag-inducing appendage that generates aerodynamic torque proportional to rotational velocity, providing passive and actively regulated stabilisation during flight without reliance on high-inertia reaction wheels. A unified dynamic model of the hybrid flight–stance system is developed, capturing the coupling between aerodynamic torques, compliant leg actuation, and underactuated body dynamics. Control strategies are formulated to regulate attitude and forward velocity through coordinated stance impulse shaping and flight-phase drag modulation. Stability is analysed using both linearised subspace representations and jump-to-jump variability metrics. Simulation studies demonstrate sustained forward hopping with stable attitude regulation and highlight the importance of coupling between drag magnitude and feedback gain selection. A Monte Carlo sensitivity analysis further reveals that the closed-loop system exhibits a finite yet substantial region of attraction under bounded parameter uncertainty, consistent with the behaviour of nonlinear hybrid systems. The results establish that combining compliant energy storage with aerodynamic drag-based balancing provides an effective embodied intelligence strategy for agile robotic locomotion. The proposed framework offers a pathway toward energy-efficient, mechanically adaptive jumping robots and contributes to the broader understanding of bioinspired actuation and hybrid dynamic stability.