Indoor Scene Reconstruction And Robot Navigation Based On 3D Laser Ranging

Under the condition of weak indoor GPS signal,mobile robot's active perception of unknown environment and autonomous navigation has been the hotspot in the field of robot research.The accurate 3D scene model is helpful for the robot to obtain its own accurate pose estimation through data matching and so on,and can help generate collision-free planning path based on the environment information.Therefore,the 3D scene reconstruction technology and the 3D scene-based navigation algorithm have a very important development significance of ruins rescue,environmental exploration and so on.In this paper,we focus on 3D scene reconstruction and path planning.After filtering and 3-D calibration of the point cloud,the mobile robot can obtain the laser point cloud of the current position in real time based on the self-developed 3D laser ranging system.In the process of moving,the robot extracts the feature points of each laser line,and then searches for the matching points corresponding to each feature point in the previous point cloud,and performs the nonlinear iteration based on the relation of the corresponding points in order to estimate the robot position.The corrected robot position is further calculated by increasing the number of feature point matching pairs.With accumulating the point cloud with fusion of robot pose,the robot can complete 3D reconstruction of the scene in real-time.Then,the reconstructed 3D scene is modeled using an octree data structure.According to the octree map,the mobile robot uses the wavefront propagation algorithm to explore in the scene,detect the collision effectively,and generate a feasible path.And then in the process of moving,the robot updates the path constantly,so as to remove the interference of dynamic obstacles to ensure the safety of mobile robot autonomous navigation.In addition,this paper presents a recognition method of the navigation logo,so that mobile robots can detect the navigation logo information in the process of self-exploration and perform the navigation task.The method of projection transformation is used to avoid distortion by Gaussian kernel filtering method,so that the information of navigation logo can be identified correctly,so as to help the mobile robot navigation.With the Pioneer 3-DX indoor robot as the mobile platform,the ROS(Robot Operating System)is used as the software platform to build the experimental platform.The experimental results verify the practicability and effectiveness of the 3D scene reconstruction and autonomous navigation algorithm.