Mobile Robot Simultaneous Localization And Mapping

Mobile robot is an important branch of robotics, is a comprehensive intelligent control system integrated many functions, such as environmental awareness, the dynamic planning and decision making, behavior control and execution. Based on the Voyager-IIA mobile robot and Visual C++platform, this paper analyses and rectifies the error which is from the movement of mobile robot, and researches on SLAM (Simultaneous Localization and Mapping) in the unknown environment. Purpose is to establish a relatively complete system makes mobile robot has the ability to independently build map in completely unknown environment.First, this dissertation measures the motion errors through a series of outdoor experiments, adopts widely-used UMBmark checking method, compiles a experimental program with parameters after-checked. After the same experiments, this dissertation finds that the motion errors have been diminished. This result proves that the systematic errors have been controlled well, and the mobile robot has the ability to relatively locate preferably.Second, use probability grid map to build local map. Calculate and fresh probability model with Bayes theorem, this dissertation compiles a dialogue to reflect surroundings data around the mobile robot. This dissertation divides the surroundings into three parts:the unknown zone described with gray; the empty zone described with green; the occupied zone described with red. Through several local mapping tests, the local map in this dissertation could reflect surroundings data preferably, and has ability to fresh itself.The biggest problem of probability grid map is that its saved information will be increased significantly as mobile robot moving. The saved information of geometric feature map is not very large. Therefore, this dissertation chooses geometric feature map to build global map. Fresh the location data with the correction for motion errors. For sensor data, this dissertation adopts double fitting approach to minimize sensor data errors. Aim at corridor, this dissertation affords distinguishing method to doors. Through the global mapping tests in Mining Floor, the global map in this dissertation could build global map autonomously when satisfying some mapping condition. The global map could describe Environment characteristics around mobile robot path visualized.Though building the local map, this dissertation describes surroundings very well, realizes real-time of local map. During building the global map, this dissertation integrates the location data and sensor model.In addition, this dissertation realizes simultaneous localization and mapping of mobile robot, as building the glabal map.