| Source searching is a common problem faced by nature and human society.In this problem,searchers need to use acoustic,optical,electromagnetic,chemical and other signals released by the source to determine the location of the source.To solve this problem,researchers have proposed and developed various autonomous source searching(ASS)methods based on the mobile sensor.The ASS method instructs the mobile sensor to sense the signal released by the source while moving towards the source according to the sensed information in the scene containing the source.However,under the weaksensing condition caused by environmental factors or the limited sensor capability,the mobile sensor cannot accurately sense the signal.The existing source searching algorithms have disadvantages such as low efficiency,high computational complexity,and inability to adapt to complex obstacle scenarios under the weak-sensing condition,which limit the application of these algorithms in practical source searching tasks.This paper aims to solve the autonomous source searching problem under the weaksensing condition.Therefore,the limitations and development trend of the current ASS method under the weak-sensing condition are sorted out firstly,then,we conduct novel studies of the modeling framework of the ASS problem under the weak-sensing condition and the source searching algorithm in obstacle-free scenes and complex obstacle scenes.Subsequently,the novel ASS method based on the convection-diffusion model,particle filter algorithm,intelligent optimization method,Bayesian method,and other related methods and theories is proposed.In particular,this paper has made a breakthrough in the design of ASS algorithms for obstacle-free scenes and complex obstacle scenes.Compared with existing ASS algorithms,the proposed methods in this paper have better performance and are more suitable for the source searching task in practice.The main work and contribution of this paper can be summarized as follows:(1)A model of ASS problem under the weak-sensing condition is established,which makes up for the deficiency of current theoretical research on ASS problem.After analyzing the ASS problem under the condition of weak perception,the object and constraint of the problem are defined and the problem model is established.Three subproblems involved in the problem are analyzed and formalized respectively,that is,the problem of forward diffusion modeling;the problem of weak-sensing process modeling and the problem of source term estimation modeling.The models established in this paper provide a theoretical framework for the subsequent algorithm research.(2)A novel ASS algorithm,namely Clutaxis,is proposed,which provides an efficient source searching algorithm in obstacle-free scenes.In this paper,the Clutaxis algorithm adapts a mechanism based on the balance between exploration and exploitation.DBSCAN clustering algorithm or genetic algorithm was used to extract the belief source are from the source term estimation and guide the mobile sensor to explore the belief source area.The mechanism has low computational complexity and the sensor can be moved towards any direction.Therefore,the Clutaxis algorithm is more flexible and efficient than existing algorithms and can achieve better source searching performance in obstacle-free scenes.(3)A novel ASS algorithm,namely intermittent cognitive strategy(ICS),is proposed,which provides an efficient source searching algorithm in scenes with road network constraint.The scene with road network constraint is a kind of complex obstacle scene in which the mobile sensor can only move along the existing road.This paper combines the intermittent search strategy with the cognitive source search algorithm to form the ICS,which regards the searching process guided by the cognitive search algorithm as the slow-sensing move phase and triggers the jump move phase when the mobile sensor is blocked by obstacles.Therefore,the mobile sensor can quickly bypass the obstacles and continue the source searching process.In the jumping move phase,the mobile sensor does not need to collect and analyze the sensed data,so the source searching process is accelerated.(4)A novel ASS algorithm,namely cognitive strategy with forbidden area(CSFA),is proposed,which provides an efficient source searching algorithm in scenes with unknown obstacles.The scene with unknown obstacles is another kind of complex obstacle scene in which the movement of the mobile sensor is restricted by various obstacles.The location and size of obstacles cannot be obtained in advance.Therefore,it cannot bypass obstacles or jump out of trap areas formed by obstacles through active path planning.In this paper,by marking some explored areas as forbidden areas,the mobile sensor can move away passively from the area obstructed by obstacles and continue to perform the source search task.(5)A prototype system of ASS under the weak-sensing condition is established,which can provide hardware and software support for real experiments and actual source search tasks.In order to meet the requirements of ASS task under weak-sensing condition,a hardware and software prototype system is constructed in this paper.The system includes mobile sensor prototype system based on an unmanned vehicle platform and ASS simulation prototype system.The prototype system constructed in this paper can be used for real experiments of source searching,so as to further verify the effectiveness of various source searching algorithms,and optimize the algorithms according to the experimental results.In addition,the ASS prototype system can assist the source searching task in practice.This paper focuses on the study of ASS method under the weak-sensing condition.In addition to enriching the modeling theory of the ASS problem,this paper carried out innovative research on the ASS problem in obstacle-free scenes and complex obstacle scenes,and proposed novel ASS algorithms that exceed the existing algorithms in terms of performance.Besides,a software and hardware prototype system for the source searching task under the weak-sensing condition was developed.This paper can promote the understanding of people in source searching,and the contributions of this research are of great significance to rescue search,source localization,nuclear-biochemical hazard source search,emergency management,and other fields. |
