Mobile Robot Based Gas Leakage Source Location And Concentration Map Construction

With the continuous development of industrialization,harmful gas leakage has become a safety problem that will endanger people’s production and life.Aiming at the application problems of mobile robots in leakage scenes,this paper proposes a comprehensive maximum location method based on Gaussian process to help robots efficiently locate gas leakage sources,and a new D-S(Dempster-Shafer)evidence theory improvement method to improve the trust of multi-evidence body fusion results.A gas information sampling method based on Gaussian process and A* algorithm is used to help the robot construct gas concentration map efficiently.The main content of this paper is as follows:1)Research on locating method of gas leakage source: A comprehensive maximum positioning strategy based on Gaussian process is proposed.Gaussian process has the ability to estimate the concentration of unmeasured positions through the observed values and endow uncertainty,which provides the search direction for the robot.Since Gaussian process prediction does not depend on obvious gas concentration gradient,It solves the problem that traditional chemotaxis,windtaxis and other strategies cannot effectively locate the leakage source in the leakage scene where the gas concentration gradient is not obvious and the wind speed and direction are unstable.2)Research on Improved Method of D-S evidence Theory: A new improvement strategy of evidence theory is proposed.Firstly,the similarity between different evidence bodies is measured by calculating the distance between different evidence bodies,so as to distinguish abnormal evidence bodies.Then,the external weight and internal weight of each body of evidence are set according to the trust of the focal element of evidence and the degree of trust,and the weighted average algorithm is used to construct new replacement evidence.Finally,the original combination rules are used to fuse the multi-source information.The experimental results show that the improved strategy proposed by standard is effective and can solve the paradox problem in the fusion of multiple evidence bodies.Compared with other similar improved algorithms,it has a higher degree of evidence trust.3)Research on construction method of gas concentration map: In order to improve the efficiency of sampling efficiency of the mobile robot in the mapping of gas distribution,this paper presents a information sampling strategy based on Gaussian process and A* algorithm,which can achieve higher sampling efficiency with smaller path cost.In each sampling process,firstly,using the uncertainty of Gaussian process prediction to obtain the position with high information entropy.Secondly,A* algorithm is used to plan the path of these high information entropy locations.Thirdly,using the corresponding average information entropy and the path cost construct the information sampling utility function,the key sampling point and the sampling path are selected by maximizing the utility function.Finaly complete the sampling and update the mapping of gas distribution,until the accuracy(Root Mean Square Error)is the standard.In the case of the same mapping of gas distribution precision,the sampling time is evaluated by comparing the sampling time of different strategies.The experimental results show that the sampling strategy of the algorithm of GP and A* algorithm has high sampling efficiency,and the efficiency is 160.0% of lane shape,133.3% of greed based on information entropy,and 106.7% of transplanted RRT* algorithm,respectively.