Research On Progressive Environment Cognition And Navigation Of Intelligent Mobile Robot

Mobile robot navigation is an important project in robot research.Nowdays,mobile robot and human beings' life is increasingly close,so it plays an important role in our life.Accurate environmental cognition and precise positioning is the foundation of the mobile robot autonomous navigation.The simultaneous localization and map building is to solve the problem of mobile robot localization and mapping.Aiming at the problem of simultaneous localization and map building problem,data association problem is studied in this paper.Maenwhile,the problem of autonomous environment exploration problem is studied,which complete the exploration and mapping tasks through fast frontier detection algorithm.In addition to positioning and map construction,the autonomous navigation of mobile robots requires autonomous motion planning to complete specific navigation tasks.Aiming at this problem,this paper studies the autonomous path planning of mobile robot based on double dqn.The main research contents are as follows:1.Data Association is the premise and basis of state estimation in both positioning and map construction of mobile robot.In this chapter,an optimized data association algorithm based on gaussian mixture clustering is proposed to solve the problems of association mismatching and high computational complexity in jointly branch and bound data association algorithm.Firstly,the association region is determined according to the sensing ability of the sensors carried by the robot.Then gaussian mixture clustering is used to group the observation features.Finally,association matching is carried out with the method of misassociation matching optimization.2.Aming at mobile robot autonomous exploration problem,we propose a mobile robot environment exploration method based on fast frontier detection.Firstly,grid map is constructed by extracting and processing the environmental information sensored by the robot.Then the fast frontier detection algorithm is used to detect the existing frontier points in the environment and extract the candidate frontier points.Selecting the best candidate frontier considering the cost of the navigation path and the information gain of the candidate frontier.Finally,the robot move to the best candidate frontier according to the A * algorithm and update the grid map.The robot repeats this process until it completes its exploration task and builds a complete raster map of the environment.3.Aming at mobile robot motion planning problem,we propose a mobile robot motion planning algorithm based on double dqn.The algorithm consists of two parts: the environment information perception processing layer and the action decision control layer.First,the robot obtain the grid data the sensor data.And process these data according to the deep neural data.Besides,make the action decision and control the robot to complete the corresponding action,and get the reward in the process of state transition.Finally,update the network parameters and optimize the actions to complete the navigation task.