Mobile Robot Autonomous Exploration And Cognition In Indoor Environment

Mobile robot is an important research field of robot technology and its research involves artificial intelligence, sensor technology, computer vision, digital image processing, etc. In many cases, the mobile robot lacks priori maps. Consequently, the mobile robot should have the ability of autonomous environment exploration. How to further autonomously complete environment cognition after accomplishing environment exploration is a research focus. This paper researches on mobile robot autonomous indoor exploration and cognition.This paper gives the environment representation based on opening-points. And then proposes the exploration strategy based on exploration-direction-points (EDPs) using the perceived information of 2D laser. Due to improve the efficiency of exploration, the strategy used in this paper considers three features to characterize an exploring target: the traveling cost, the distance between the mobile robot and the information gain. And utility function is used to evaluate each EDP to make the robot select the optimal EDP as target of next step.Derives from region division and overall cognition for indoor environment, this paper proposed environment segmentation approach based on environment opening-points. In the course of exploration, the environment was segmented to different regions and the opening-points were also extracted. Then connect every opening-point as tree structure to build the topological opening-points map which indicates spatial structure of every adjacent region in the global environment. During the course of real-time exploration, the topological opening-points map will be updated real-time.This paper analyzed and organized the feature information in feature invariance way based on the topological opening-points map. Then the environment marks model based on opening-points was proposed. And autonomous environment cognition was completed by environment marks matching. Simulation results and experiment results implemented in the Pioneer3-DX mobile robot and experiment data analysis show the proposed approaches' validity and practicability.