This paper presents a novel hierarchical control architecture for cooperative control of multiple tugboats to transport a large vessel. To this end, we formulate dynamic interactions between a vessel and tugboats leveraging so-called Udwadia–Kalaba equation, resulting in a highly nonlinear system model. The high nonlinearity and relatively high dimensionality make it challenging to directly apply a promising constrained control methodology, nonlinear model predictive control, due to the large computational burden. To address the issue, we propose a two-layer control architecture that combines a nonlinear model predictive controller with a simplified model at the higher layer and an optimization-based controller that guarantees constraints on interaction forces at the lower layer. The controller is finally verified through numerical simulations, where the present hierarchical controller is shown to significantly reduce the computational load while ensuring constraint fulfillment.