Attainable-Set Model Predictive Control for AUV Formation Control

Abstract

In this article, we focus on the motion control of an AUV formation in order to track a given path along which data will be gathered. A computationally efficient architecture enables the conciliation of onboard resources optimization with state feedback control-to deal with the typical a priori high uncertainty-while managing the formation with a low computational and power budgets. To meet these very strict requirements, a novel Model Predictive Control (MPC) scheme is used. The key idea is to pre-compute data which is known to be time invariant for a number of likely scenarios and store it on-board in appropriate look-up tables. Then, as the mission proceeds, sampled motion sensor data, and communicated data is processed in each one of the AUVs and fed to the onboard proposed MPC scheme implemented with the dynamics of the formation that, by combining with information extracted from the pertinent on-board look-up tables, determine the best control action with inexpensive computational operations. Index Terms-Model predictive control, Attainable set, AUV formation control, Obstacle collision avoidance Characterize obstacle No obstacle in range Obst. In range Re-plan to overcome obstacle Overcoming obstacle