| Mobile Wireless Sensor Network(MWSN)is a Sensor network developed rapidly in recent decades.When an accident occurs in a coal mine,the traditional networks are usually damaged and unable to transmit downhole information,which is not conducive to subsequent rescue work.However,the mobile wireless sensor network,with its self-repairing feature,provides a possibility to solve this problem.Aiming at the problems of node positioning in the early stage of mobile sensor network in coal mine,the robot navigation and the establishment of environmental map in unknown environment after disaster,and the task assignment of robot group with multi-repair assignment are studied in the following aspects.Firstly,aiming at the problem of wireless sensor network node location,put forward a three dimensional localization algorithm based on RSSI ranging.The determination method of 3d overlapping region and the calculation method of probability density integral of overlapping region in the algorithm are mainly solved.And it is verified by the simulation that the algorithm can provide random arrangement of mobile robot nodes with relatively accurate initial global position,which makes the subsequent reorientation of robot that relys on environment map can be well matched to the corresponding position of the map.Secondly,a method based on Visual Simultaneous Localization and mapping(VSLAM)is introduced for network destruction caused by coal mine disasters and accidents.The methods of pose calculation of mobile robot are studied,and the advantages and disadvantages of several mainstream positioning methods are compared.At aspect of map feature point extraction,realize the optimization of feature point by extracting total number of corners,realize the stable extraction of the feature points under different shooting distance through corner detection under different resolution,realize rotation description of corner points by defining the direction of the image block.And on these basis,the accuracy of image matching is improved by selective preference of the feature point.At aspect of robot pose estimation,the latest nonlinear optimization method is adopted,and the final effect of robot pose estimation is improved by state estimation of observation data with noisy collected at all times.Thirdly,the reconstruction of dense map based on vision is studied,and the reconstruction effects of several different schemes are compared.At the same time,combining with the application requirements of the repair of the incomplete network in the underground coal mine,this paper focuses on the navigation of the robot in the process of map reconstruction.The navigation based on fuzzy algorithm enables the mapping robot to find a route from the initial position to the designated point at the outer end of the roadway by avoiding obstacles,and at the same time,it completes the map establishment for repairing the network after the disaster.In the study of multi-task assignment in multi-mobile robot group,assignment algorithm with the balanced load is used to achieve the basic balance of single robot task in multi-mobile robot group.In the last part,field experiments of the robot are carried out to verify the previous researchs of visual positioning and mapping.In a simulated environment with artificial lighting in the corridor,the mapping robot detected three preset target points from the initial position.The running path of robot did not collide with the obstacles and reached the target point smoothly with the specified pose.Moreover,the mapping was basically accurate in describing the environmental information.In a word,these all achieved the desired effect. |
PDF Full Text Request