Intelligent Optimization And Dynamic Coordination Of Multi-robot Patrolling System

Multi-robot systems are considered to be promising in the field of security surveillance and disaster response due to their distributed parallel processing abilities.As one of the most important work in the concerned applications,a good patrolling strategy is not only needed to be optimized from a global perspective but also can possess robots of autonomous cooperative abilities for dynamic environments.In view of the above-mentioned demand,this dissertation studies centralized and distributed multi-robot patrolling systems concerning different applications from simple and static to complex and dynamic.It first utilizes bio-intelligent inspired algorithms to design globally optimal patrolling paths for robots and then proposes cooperative strategies based on event-driven control to enhance their adaptiveness to dynamic environments and improve working efficiency.The major contributions of this dissertation are as follows.(1)Since multi-robot patrolling was first comprehensively addressed in 2002,a number of research achievements overcoming different difficulties and focusing on various key points are made.According to different patrolling goals,this dissertation reviews the existing works from the perspectives of regular and adversarial patrolling.Regular patrolling requires robots to visit important locations as frequently as possible and a series of deterministic strategies are proposed,while adversarial one focuses on unpredictable robots' moving patterns to maximize adversary detection probability.Under each category,a systematic survey is done including problem statements and modeling,patrolling objectives and evaluation criteria,and representative patrolling strategies and approaches.Open questions are presented accordingly.(2)A basic patrolling problem can be seen as a multi-robot multi-task allocation problem.This dissertation proposes a niching immune-based optimization algorithm based on Softmax regression for it to handle related multimodal optimization problems.A pre-judgment of population is done before entering an evaluation process to reduce the evaluation time and to avoid unnecessary computation.Furthermore,a guiding mutation operator inspired by the base pair in theory of gene mutation is introduced to strengthen its search ability.When a gene mutates,others in the same gene group may mutate with a higher probability.The effectiveness and efficiency of the proposed algorithms is verified through a series of experiments on benchmark functions and two multi-robot task allocation cases.(3)For the problem that multi-robot systems are required to patrol a concerned part of ocean to ensure maritime safety under severe weather conditions,this dissertation builds a multi-objective model with the consideration of resource,timeliness and emergency simultaneously and presents a novel approach inspired by an immune-endocrine short feedback system to solve it.Regulations produced by an endocrine system act on two phases of an artificial immune algorithm.First,a kind of hormone applied in a mutation process is proposed to decrease the number of undesirable solutions with the help of a Bayesian learning.Second,it is performed at a memory cell to suppress high-concentration antibodies and save high-quality individuals.Experimental results illustrate the desired effects of two regulation phases and the proposed method's performance in solving a maritime patrolling problem.Its high search ability and convergence speed are shown via its comparison with the other well-known algorithms.(4)For a dynamic patrolling problem caused by the movements of objects,this dissertation proposes a distributed event-driven cooperative strategy for multi-robot systems to achieve the coordination among robots autonomously.The designed forward and backward utility functions provide a two-way evaluation for a robot to choose its target with consideration of the states of other robots.The reasonable transfer of targets among robots occurs according to a defined cooperative action.The released robots can visit more objects,which leads to an improved working efficiency and patrolling frequency.Simulation results show that the proposed strategy can increase visit frequency of objects.The Unmanned Surface Vehicles(USVs)executing this strategy are able to collect measurement data of both liquid and gaseous contaminants via embedding sensors in a real time.Finally,conclusions are drawn in this dissertation.The inadequacies of current works and the promising studies are discussed for a further improvement.