Centralised versus Decentralised Control Reconfiguration for Collaborating Underwater Robots

Abstract

The present paper introduces an approach to fault-tolerant reconfiguration for collaborating underwater robots. Fault-tolerant reconfiguration is obtained using the virtual actuator approach, Steffen (2005). The paper investigates properties of a centralised versus a decentralised implementation and assesses the capabilities under communication constraints between the individual robots. In the centralised case, each robot sends information related to its own status to a unique virtual actuator that computes the necessary reconfiguration. In the decentralised case, each robot is equipped with its own virtual actuator that is able to accommodate both local faults and faults within a collaborating unit. The paper discusses how this is done through exploiting structural information (e.g. thruster configuration) for each participant in the cooperation. A test scenario is presented as a case in which an underwater drill needs to be transported and positioned by three collaborating robots as part of an underwater autonomous operation.

Game Theory Approach for Communication Security and Safety Assurance in Cyber-physical Systems with Reputation and Trust-based Mechanisms

“…A more detailed review of their properties and an example of practical application can be found in [5,6]. In the present work, we use a decentralized agent control strategy, as it is more reliable and fault-tolerant from the safety perspective.…”

A framework for cyber-physical production system management and digital twin feedback monitoring for fast failure recovery

“…On the one hand, centralized methods are the most common, and they make a decision through all the information about the systems, being the drawback that fast communication methods and massive data elaboration are needed. On the other hand, decentralization allows each agent to process its information locally but lacks a complete vision of the whole environment (Furno et al 2015;Son 2011).…”

Pose Consensus of Multiple Robots with Time-Delays Using Neural Networks

“…[1][2][3] The objective in cooperative control is to design a distributed control strategy using only local information, so that the aggregate system achieves specified behaviors, such as flocking, [4,5] synchronization, [6,7] coordination, [8,9] rendezvous, and formation control. The present and future practical applications of multi-robot systems span different areas, such as: underwater and space exploration (underwater cultural heritage recovery, coordination of clusters of satellites, and synchronization of spacecrafts); [10,11] hazardous environments (search and rescue missions, and military operations); [12,13] and service robotics (commercial cleaning, material handling, furniture assembly, etc.). [14][15][16] A fundamental approach to achieve a cooperative behavior is consensus.…”