C3D: Cascade Control with Change Point Detection and Deep Koopman Learning for Autonomous Surface Vehicles

TL;DR

A data-driven approach based on modular design for ease of transfer of autonomy across different maritime surface vessel platforms alleviates issues related to a priori identification of system models that may become deficient under evolving system behaviors or shifting, unanticipated, environmental influences.

Abstract

In this paper, we discuss the development and deployment of a robust autonomous system capable of performing various tasks in the maritime domain under unknown dynamic conditions. We investigate a data-driven approach based on modular design for ease of transfer of autonomy across different maritime surface vessel platforms. The data-driven approach alleviates issues related to a priori identification of system models that may become deficient under evolving system behaviors or shifting, unanticipated, environmental influences. Our proposed learning-based platform comprises a deep Koopman system model and a change point detector that provides guidance on domain shifts prompting relearning under severe exogenous and endogenous perturbations. Motion control of the autonomous system is achieved via an optimal controller design. The Koopman linearized model naturally lends itself to a linear-quadratic regulator (LQR) control design. We propose the C3D control architecture Cascade Control with Change Point Detection and Deep Koopman Learning. The framework is verified in station keeping task on an ASV in both simulation and real experiments. The approach achieved at least 13.9 percent improvement in mean distance error in all test cases compared to the methods that do not consider system changes.

Figures (13)

• Figure 1: WAM-V 16 ASV with an unbalance load performing station keeping in Lake Harner, IN
• Figure 2: System architecture. At every time step, the vehicle sends the current state $s_t$ and control command $a_t$ to the change point detector, deep Kooman learner, and cascade controller. The inner loop of the cascade controller uses the newest matrices $A_\tau, B_\tau, C_\tau$ and Koopman operator $g(\cdot,\theta_\tau)$ to track the target velocity $\nu_k^*$ generated by the outer loop.
• Figure 3: Step input response and bode plot of identified linear system
• Figure 4: Example of the designed random walk signal (top) and the Fourier transform (bottom)
• Figure 5: Diagram of the vehicle, the target, and distance to the virtual line perpendicular to the yaw angle