Research On The Method Of Robot Plume Tracking In 3D Environments

While the rapid development of industrial society brings convenience to people,serious safety accidents have been occuring frequently caused by the leakage of flammable and explosive toxic gases.To accurately provide the location of pollution sources in a short time,effectively control the scope of pollution,make early warning and personnel evacuation in time,and minimize economic and property losses,how to locate the leakage source of flammable and explosive toxic gases has become an urgent problem that to be solved in the field of public safety.The gas propagates in the form of plumes in mediums.A large number of facts show that organisms can use olfactory and(or)visual information to achieve a variety of social activities,such as foraging,communication,courtship,enemy defense,and so on.Inspired by biological behaviors,information such as smell and/or vision are used for mobile robots to locate plume sources.At present,the researches on plume tracking and plume source location are mainly limited to the two-dimensional space,which means that the olfactory sensor can only detect the plume in a specific plane.Olfactory sensors and wind speed sensors are used by many researchers to study the location of plume sources.Although good results have been achieved,but for the objective law of plume distribution in 3D environments been ignored,those methods may be completely ineffective in some cases.In view of the lack of researches on plume tracking and plume source location in 3D environments,the implementation mechanism of Gray Wolf Optimizer is analyzed and a robotic plume tracking method based on Gray Wolf Optimizer is proposed in this paper.Considering that the wind direction and wind speed are not significant in the weak wind or diffusion environments,the use of wind direction and wind speed sensors are canceled in this paper.The plume is tracked and the plume source is determined by simulating the social mechanism and hunting behavior of the gray wolf population by the robot.Compared with the Z-shaped traversal algorithm,Particle Swarm Optimization algorithm,and Genetic Algorithm,the feasibility and effectiveness of Grey Wolf Optimizer to solve the problem of plume tracking and plume source location is verified.In the four groups of simulation experiments,the average running time of plume source location is 65 s,135s,140 s,65s,respectively.The average driving distance is 12.85 m,20.12 m,25.37 m,13.62 m,respectively.The success rate of location are 92%,94%,94%,94%,respectively.To verify whether the gray wolf optimization algorithm is still effective in solving the actual plume tracking problem,three groups of experiments are designed in this paper.In three practical experiments,the average running time of the plume source location is 143.5s,153.9s,and 125.3s,respectively.The average driving distance is 12.36 m,11.19 m,12.25 m,respectively.The success rate of location is 95%,90%,and 90%,respectively.The effectiveness of Grey Wolf Optimizer to solve the actual plume tracking problem is verified.To improve the success rate and speed of locating plume source,a plume tracking method based on visual and olfactory information fusion is proposed in this paper.Considering the existence of suspicious plume sources,the visual-olfactory information fusion method is used to track and identify the plume source.To improve the speed of plume search and the reliability of early plume tracking,the robot tracks the plume by simulating the social mechanism and hunting behavior of the gray wolf population.In the plume source confirmation,the visual and olfactory data are fused to improve decisionmaking and the system's reliability through the weighted fusion function.A technical strategy is designed to enable the autonomous mobile robot to complete the plume source location task in this paper.The subsumption architecture is used to define and arbitrate behavior priority to coordinate different behaviors.In the experiment,the average positioning error is 0.13 m,the average running time is 147.7s,and the average stroke is21.35 m.The experimental results show that the plume tracking method based on visual and olfactory information fusion can well realize the plume source location's task in 3D environments.