Event-Based Adaptive Koopman Framework for Optic Flow-Guided Landing on Moving Platforms
Bazeela Banday, Chandan Kumar Sah, Jishnu Keshavan
TL;DR
The paper tackles robust vertical UAV landing on moving platforms using only monocular optic flow by learning a global linear Koopman model of the optic flow-to-thrust dynamics and refining it online to compensate for ground effects and platform motion. It combines offline Extended Dynamic Mode Decomposition with an online adaptation law and integrates this with an event-triggered Model Predictive Control framework, solved via OSQP, to regulate optic flow toward a reference while enforcing constraints. The main contributions include the first integration of event-triggered adaptation within a Koopman-based framework, formal convergence guarantees to a uniform ultimate bound with Zeno-free operation, and simulation evidence showing robust landings under noise and platform oscillations with substantial reductions in update events and computation. The work promises a computationally efficient, vision-based strategy for soft landings on dynamic platforms, enabling practical deployment on small UAVs with limited sensing and processing capabilities.
Abstract
This paper presents an optic flow-guided approach for achieving soft landings by resource-constrained unmanned aerial vehicles (UAVs) on dynamic platforms. An offline data-driven linear model based on Koopman operator theory is developed to describe the underlying (nonlinear) dynamics of optic flow output obtained from a single monocular camera that maps to vehicle acceleration as the control input. Moreover, a novel adaptation scheme within the Koopman framework is introduced online to handle uncertainties such as unknown platform motion and ground effect, which exert a significant influence during the terminal stage of the descent process. Further, to minimize computational overhead, an event-based adaptation trigger is incorporated into an event-driven Model Predictive Control (MPC) strategy to regulate optic flow and track a desired reference. A detailed convergence analysis ensures global convergence of the tracking error to a uniform ultimate bound while ensuring Zeno-free behavior. Simulation results demonstrate the algorithm's robustness and effectiveness in landing on dynamic platforms under ground effect and sensor noise, which compares favorably to non-adaptive event-triggered and time-triggered adaptive schemes.