Research On Method For Discontinuous Target Tracking With Autonomous Underwater Vehicles

The research on olfactory biomimetic robots has become international focus since 1990s.That is to sense,search and track the plumes in water or air which diffused by chemicals or other substances,and then locate the source location by imitating biological olfactory.Robots equipped with chemical sensors to percept,search,track the chemical plumes and ultimately locate the source locations precisely,the potential applications for them include the environmental monitoring,chemical plants safety,searching and rescuing,counter-terrorisms,narcotics control,explosive and archaeological clearance and submarine hydrothermal vents positioning.If the autonomous underwater vehicle(AUV)endows with olfactory perceptive ability,tracks the chemical plumes and locates the source locations of the plumes or the hydrothermal vents in deep-sea,the research will have broad prospects and great significance.As a single AUV often suffers from its size,power it carried,limited sensors categories and numbers installed on it,using it to trace chemical plume will often lead to relapsing into the minimum value of the local gradient field and locating the source location difficultly,which result in failure to the chemical plume tracing(CPT)process.While the distributed system of multiple AUVs will become the first choice for the chemical plume exploration(CPE)mission because of its high-precision navigational and positioning ability,adaptive measures capacity for its temporal and spatial distribution,autonomous planning for the complex marine environments,and the coordinated and cooperated ability.It is defined and studied a class of discontinuous target in the paper.Discontinuous target may be hydrothermal plumes,Mine emissions,signal of black box of aircraft,pinger.Performing improved researching works for the CPE missions that via a single AUV based on the mothinspired behavioral strategies developed by Wei Li,et al.We optimize the behavioral parameters for the Find-Plume behavior and Maintain-Plume behavior using Monte Carlo simulation approach.We develop a simulation system for the CPE via a single AUV,which based on the plume model that developed by Wei Li et al,the optimized behavioral parameters and the vehicle's parameters.We implement two source identification algorithms of SIZ_T and SIZ_F that developed by Wei Li in the CPE simulation system.We keep changing the chemical plume in a fluid environment over 130 hours and perform 1000 CPT test runs for both algorithms.Simulation results show the effectiveness of these two source identification algorithms.We concern the CPT time cost and source identification errors and make comparisons among these two algorithms,also we make comparisons between the simulation results obtained from this thesis and the results proposed by Wei Li in his literature.Then we can draw a conclusion that the SIZ F algorithm is superior to SIZ T algorithm,and the CPT performance in this thesis has been improved significantly.It is proposed the method that uses gradient direction by studying maintain behavior in target in bionic behavior to approach source of target.In the same time,it is proposed the method that uses multi-objective optimization to improve the method of gradient direction and avoid sensitivity of single sensor.At last,we use the simulation environment of Opengl and C++ to confirm the superiority of whole behavior method and the validity of the paper.The paper provides a program for discontinuous target tracking.