Design Of Detection Robot Localization System Based On Multi-Sensor Fusion In Flood Discharge Tunnel

The structure integrity and flood carrying capacity of tailings pond flood discharge tunnel is directly connected with the safety of the mine,the flood discharge tunnel has characteristics of high environmental repeatability,lack of feature points,darkness and have waterflow etc.Therefore,the current mainstream positioning method has poor applicability in this scene,and is hard to realize precise positioning of remote distance.In order to solve above problem,this dissertation conducts research in the following aspects.Firstly,analyzes by aiming at the real environment and engineering project demand of tailings pond flood discharge tunnel,and then carries out overall framework design of positioning system of detection robotic(hereafter referred as“robot”)of flood discharge tunnel,divides the system into four parts according to functional difference,and combines known parameter to carry out scheme analysis and equipment model selection for various sub-systems.Adopts Ar Uco code as road sign,and proves it has the advantage of strong robustness to illumination through experiment,it applies to tailings pond flood discharge tunnel environment that has no natural light and needs the robot to carry illuminant by itself.Secondly,establishes speed motion model and odometer motion model,carries out simulation experiment aiming at current existing 3 kinds of odometer motion model,result indicates that secant motion model has high precision and good stability.The UMBmark odometer demarcate algorithm implementation principle is introduced.Establishes inertial measurement unit model and analyzes positioning principle.Establishes camera model and demarcates,analyzes camera pose conversion principle.Then,analyzes specific procedure and implementation principle of multi-sensor data fusion localization algorithm.Aiming at the problem of difficulty in positioning of the tailings pond flood discharge tunnel environment,takes the pre-laid Ar Uco code as road sign,takes Ar Uco code as boundary for far distance positioning,segments into multiple sections of short distance positioning.The algorithm uses loose combination extended Kalman filter to combine odometer with inertial navigation,and realizes short distance precision positioning,and verify the correctness of the model through emulation;uses graph optimization method to conduct optimal pose estimation near the Ar Uco code,and combines the absolute pose information carried by Ar Uco code to realize accumulative error correction,and completes autonomous positioning of robot.Finally,uses the robot system designed and established by this dissertation to carry out experiment to verify the correctness of the model and algorithm of this dissertation,and carries out real place experiment in the tailings pond flood discharge tunnel environment,result indicates that the robot system applies to this environment,and the multi-sensor fusion localization algorithm has high stability and precision,and could effectively correct accumulative error and realize precise positioning of robot in the flood discharge tunnel environment.The average error after rectification at the Ar Uco code is 4.2 mm,calculation takes corrected positioning information as theoretical value gets,the average positioning error at(0 m,20 m)is 0.198 m,the average positioning error at(0 m,40 m)is 0.212 m,.