The containment maneuvering of marine surface vehicles has two objectives. The first one is to force the marine vehicles to follow a convex hull spanned by multiple parameterized paths. The second one is to meet the requirement of a desired dynamic behavior along multiple paths during containment. A modular design approach to the containment maneuvering of marine surface vehicles is presented. At first, an estimator module using a recurrent neural network is proposed to estimate the unknown kinetics induced by model uncertainty, unmodeled dynamics, and environmental disturbances. Next, a controller module is developed based on a distributed path maneuvering design and a linear tracking differentiator. Finally, two path update laws based on a maneuvering error feedback and a filtering update scheme, respectively, are constructed. The estimator-controller pair forms a cascade system, which is proved to be input-To-state stable. The developed controller has a desirable spatial-Temporal decoupling property, and geometric and dynamic objectives can be achieved separately. Results of comparative studies are provided to substantiate the efficacy of the proposed method.
- Containment maneuvering
- dynamic surface control (DSC)
- recurrent neural network (RNN)
- spatial-Temporal decoupling
- tracking differentiator (TD)