A Time and Place to Land: Online Learning-Based Distributed MPC for Multirotor Landing on Surface Vessel in Waves

Abstract

Landing a multirotor unmanned aerial vehicle (UAV) on an uncrewed surface vessel (USV) extends the operational range and offers recharging capabilities for maritime and limnology applications, such as search-and-rescue and environmental monitoring. However, autonomous UAV landings on USVs are challenging due to the unpredictable tilt and motion of the vessel caused by waves. This movement introduces spatial and temporal uncertainties, complicating safe, precise landings. Existing autonomous landing techniques on unmanned ground vehicles (UGVs) rely on shared state information, often causing time delays due to communication limits. This paper introduces a learning-based distributed Model Predictive Control (MPC) framework for autonomous UAV landings on USVs in wave-like conditions. Each vehicle's MPC optimizes for an artificial goal and input, sharing only the goal with the other vehicle. These goals are penalized by coupling and platform tilt costs, learned as a Gaussian Process (GP). We validate our framework in comprehensive indoor experiments using a custom-designed platform attached to a UGV to simulate USV tilting motion. Our approach achieves a 53% increase in landing success compared to an approach that neglects the impact of tilt motion on landing.

Figures (5)

• Figure 1: Experimental setup showing (1) the Bitcraze Crazyflie 2.1 multirotor and (2) ClearPath Robotics Husky unmanned ground vehicle (UGV) equipped with (3) a custom-designed tilting platform. The multirotor communicates with the UGV for cooperative landing. The platform emulates the tilting motion of a USV in waves, enabling experimental validation of the proposed distributed model predictive control (MPC) framework.
• Figure 2: Overview of our custom tilting platform for ground vehicles. In (a), the platform is shown mounted on the deck of the ClearPath Robotics Husky, (b) shows a closer view of the platform's components, and (c) highlights the ball-and-socket joint and the linkage mechanism.
• Figure 3: Experiment 1 and 2: We visualize the tilt of the platform at landing in (a) calm wave conditions and (b) harsh wave conditions. In both experiments, our proposed approach (blue) achieves a lower landing tilt at landing over the purely cooperative strategy (red). Our proposed distributed MPC scheme can locate a low-tilt landing location from all six initial platform positions.
• Figure 4: Experiment 2 visualization of (a) final $x$-$y$ landing locations, (b) landing locations on the platform, and (c) the average linear vs rotational kinetic energy of the tilting platform. Failed landings are represented by an $\times$.
• Figure 5: Experiment 3: We learn the wave model from Experiment 2 using a GP. We visualize the tilt of the platform at the landing using pure cooperation (red), a mean cost and high variance cost (yellow), and only a mean cost (blue).