Research On Dynamic Environment Perception And Mapping For Autonomous Mobile Robot

In the research field of intelligent robot, the realization of the robot's autonomy is an indispensable part. It must know the environment message and local place in real time when a robot executes a task independently,then the technology of simultaneous localization and mapping appears. An autonomous mobile robot keeps walking and percepting the environment with a sensor from which gets the data used to extract representative features that can represent the environment. On the one hand, these festures can incarnate the information of the environment to some extent. On the other hand,they can correct the pose of the mobile robot comparing with the environment. So,a research on key tichnology of SlAM is built.Firstly, it need to build robot system platform, on which sets up a complex coordinate system, with environmental feature map model of perception, robot motion system model and environmental feature measurement model.Secondly, it proposes line segment feature extraction method according to the data set of environment perception. This method contains several detail disposions: multiple independent connected regions that are obtained by the initial partition of the environment perception data set, data smoothing processing in each connected regions, connected region segmentation in detailed, all these can complish the maximum segment of the perception data; using improved least square line fitting method to obtain the linear range and angle parameters, methods can obtain line segment endpoints' coordinate parameters, such as the method of processing the noise points, the method of the intersection of the line segments, and the method of filling the small scale scanning blind areas.Thirdly, it uses ICNN data association method, building data association model and corresponding data association matrix. Inspired by the method of setting the filtering threshold,a gray area partition method for feature matching is proposed,which can filter out the impossible map feature elements in the matching range effectively and reduce the dimension of the data association matrix 's solution space. The method of judging the discrimination matching pairs and disposing plan, and the method of determining effective point features, which are proposed to administer the data association of line features. The accurate and fast matching between the observation feature set and the map feature set is effectively implemented based on the ICNN data association scheme.Finally, the construction scheme of line feature map is designed, based on the principle of EKF, which includes the detailed steps of the environment perception and composition method for an autonomous mobile robot. By using the real environment perception datas, the results of experimental verification clarify the validity of the method of line feature map construction based on EKF.