Real-time Distributed MPC for Multiple Underwater Vehicles with Limited Communication Data-rates

TL;DR

This paper introduces a real-time distributed model predictive control framework for multiple underwater vehicles that optimizes communication quantization to operate effectively under limited data-rate constraints.

Contribution

It presents a novel two-stage approach combining offline quantization design with online distributed MPC, ensuring stability and feasibility in low-bandwidth underwater vehicle control.

Findings

Effective quantization design improves communication efficiency

The control framework maintains stability under limited data rates

Simulations demonstrate reliable multi-AUV coordination

Abstract

Controlling a fleet of autonomous underwater vehicles can be challenging due to low bandwidth communication between agents. This paper proposes to address this challenge by optimizing the quantization design of the communications between agents for use in distributed algorithms. The proposed approach considers two stages for the problem of multi-AUV control: an off-line stage where the quantization design is optimized; and an on-line stage based on a distributed model predictive control formulation and a distributed optimization algorithm with quantization. The standard properties of recursive feasibility and stability of the closed loop systems are analyzed, and simulations used to demonstrate the overall behaviour of the proposed approach.