Study On Key Navigation Technology Of Coal Mine Exploration Robot

Navigation environment is analyzed of coal mine robotic explorer, and therunning gear and control system of which is analyzed and designed. Based on theplatform of this type o f robot, navigation environment monitoring and identificationmethod is discussed, and the navigation method is studied detailedly according toarbitration fusion coordination strategy. Afterwards, the conclusion with significanceis obtained to enhance the navigation sense of coal mine robotic explorer.Firstly, topographic feature within the working face and equipment layoutunderground coal mine are analyzed. With researching the underground coal mineenvironment that the robot needs to overcome, and investigating and featureextraction of working face and roadway environment at regular production and afterdisaster, the navigation environment of coal mine robotic explorer is reduced to fourtypical terrain, and the range of sizes is provided, including boss, slope, continuousladder and ditch.Secondly, in order to increase the obstacle climbing ability of coal mine roboticexplorer, a asymmetrical ditrigon tracked suspension is designed, the homologoustheoretical model is established and the obstacle crossing character is analyzed.Through the analysis shows that the running gear of this swing asymmetrical ditrigontracked robot possesses well passive terrain adaptability and obstacle crossingperformance. In addition, the composition is discusses of robotic explorer controlsystem, the overall plan of which is designed on the basis of system function analysis.Through analysis and comparison, the PSoC integrated developing platform is used asprocessor, and the hardware circuit and software program of control system isprojected detailedly according to further prototype. Afterwards, the PCB and theproduction of material object are completed.Thirdly, the basic principles of classic kalman filter, attenuated memory filtering,limited memory filtering and the way of their combination were analyzed. A newmethod that limited memory kalman filter with exponential fading factor was putforward for coal mine rescue robot self-localization. According to the components ofself-localization system of coal mine rescue robots, self-localization based on deadreckoning was introduced. By using limited memory kalman filter with exponentialfading factor fusing multiple sensors’ data stage by stage, improved dead reckoningmethod was proposed to achieve the minimum error on the odometer of control. Hardware system based on STC12C5A60S2MCU as the core of the multi-sensorinformation fusion was established, and software algorithm was given. On basis ofthese, experiments on coal mine rescue robot self-localization were carried out tojustify the method proposed.Then, through researching some kinds of matching method and aiming at therequirements of real-time of robots, one matching method is proposed whichcombines template and feature point matching. Then, the method of gaining barrierthree-dimensional information taking advantage of single-point method of epipolarlines. Firstly, the barrier margin is cut apart and the accelerated matching is proceededmaking use of epipolar lines. Then, the matching points of each margin maximalpoints are obtained one by one, and the vidicon coordinate of each margin maximalpoint is acquired, and the three-dimensional information of barrier is obtainedincluding altitude,width and distance. After that, the mathematical model of gradientcomputation is established, and the model of estimating gradient terrain is received oninclined gradient making use of RBF method. Finally, continuous ladder image borderis detected using Canny algorithm and the calculating method of ladder deviationangle is provided after fitting a straight line with least square method.Then, in order to enhance comprehensive ability of coal mine robotic explorer tonavigation environment and on the basis of obstacle crossing ability of CUMT-5coalmine robotic explorer, the analysis and evaluation method of passing performance ofterrain are discussing taking advantage of fuzzy theory, image processing techniquesand nerual network technique. Furthermore, with the advantages of the fusionmechanism and arbitration mechanism blended in behavior coordination strategyaccording to project of behavior controller to typical environment model, thecoordination control is realized adaptively to all kinds of basic behavior. On this basis,the navigation of coal mine robotic explorer comes true ulteriorly in complexnon-structural environment with terrain passibility evaluation brought in coordinationstrategy.Finally, the experimental verification is conducted of the key technology of coalmine robotic explorer navigation.