Research On Indoor Exploration And Navigation Of Mobile Robot Based On Scanning Laser Range Finder

Environment modeling,localization and path planning are three basic tasks of robot navigation.In the absence of any environmental information,the robot should first explore the environment independently and establish an environment model,and then do follow-up tasks according to the built model,such as global localization and path planning.In order to perceive the environment to complete navigation task,laser range finder which can sense distance are widely used in navigation robots.In this paper,the environment exploration and navigation problem of robot with laser range finder in unknown indoor environment is studied.1.Aiming at the problem of robot environment exploration and modeling,a forward search algorithm based on continuous grid map is proposed.The algorithm uses radial basis function neural network to establish the corresponding local continuous grid map for the region perceived by the range sensor,and then uses the gradient field of continuous map to efficiently calculate the probability frontier map.On this basis,a meaningful frontier is selected according to a certain threshold,and the next exploration target point is extracted by k-means algorithm.In the process of building the map step by step,the bayesian committee machine(Bayesian Committee Machine)is used to integrate the local map into the global map.The simulation results show that the algorithm can make the robot generate target points selectively and improve the efficiency of environmental exploration.2.In order to solve the global localization problem based on known maps,a model matching algorithm in hough space is proposed to locate the mobile robot.In this algorithm,hough transform is introduced to the robot localization problem.When there is no initial position and pose information,the local environment data obtained by the laser scanning range sensor is matched with the known environment model in the hough space,and the global localization is obtained.First,the translation invariant function is constructed,and the rotation angle between the two data sets is obtained by translation invariance and cross correlation function.Then,according to the obtained multiple angles,the data obtained by the corresponding laser scanning range sensor are rotated,and the cross correlation function of the hough transform is constructed with the known environment model,and the local extremum of the mutual function is obtained again to obtain the possible translational amount between the local data and the known model.The possible rotational angular shift of the rotation angle obtained by the laser sensor data observed at several discontinuous time points is finally obtained by the global localization position of the mobile robot.The robot localization process is simulated using simulation environment and real data set.The results show that the algorithm can achieve the goal of global localization.3.Path reliability is one of the important factors for mobile robot to move in indoor environment.In order to ensure the reliability of the mobile robot path,a path planning algorithm based on the sampling path planning algorithm and information theory is proposed in order to ensure the reliability of the mobile robot path.On the basis of fast searching for random tree and fast searching information collection algorithm,the algorithm adds mutual information to evaluate the environmental uncertainty of the new sampling point,and directing the generation of random tree with the distance between the sampling point and the target point.Based on the known map,the algorithm generates planning paths with both reliability and mobile distance.The simulation results verify the effectiveness of the proposed method.