The Research Of Semantic Map Updates And Robot Localization Based On RFID And Laser Sensor

With the rapid development of robotics technology, s Service robots are now generally participating into our daily lives, it plays an increasingly important role in helping senior people,disabled, medical rehabilitation, public safety and other areas. Service robots usually work in indoor environments such as home and office, these environments structure are usually complex and often change. So the robot is required to complete the service tasks automatically and intelligently.RFID technology, as an new automatic identification technology, with the characteristics of non-line-of-sight identification, large data capacity, fast recognition, remote identification and so on, has been widely applied in the field of Internet of Things. The utility of RFID technology to the field of robotics can provide more environmental information for the robot to complete service tasks, so it has become a hot topic in the field of robotics research in recent years.On the fundamental achievements of the research on robot navigation at home and abroad, and with the application of RFID as the central technology, this thesis introduces the environmental modeling, map updates and robot localization. The main contributions of this thesis are shown as follows.In order to complete smart navigation of the robot driven by service task, the thesis proposes the construction method of semantic map in the indoor environment based on RFID technology and laser sensor. The semantic map is make up with global topology map, local grid map and semantic repository. In order to manage the semantic information of objects, they are divided into static objects, semi-dynamic objects and dynamic object according to the object characteristics, the RFID tags are also classified by the objects they paste on, and design the data format of RFID tags. The global topology map describe the relationship of rooms, functions and objects attribution based on RFID tags as nodes, build by spectral clustering algorithm. and design the semantic repository. The local grid map describe the environmental geometric characteristics and provide navigation information. In order to map updates, the grid map is constructed by dividing into static grid map and semi-dynamic gird maps.As the indoor environment is variable, this thesis presents a semantic map update method based on RFID technology. As for semi-dynamic grid map updates, the thesis made a detailed description on semi-dynamic object identification, linear feature extraction and the fitting line attributes. The information of object’s shape and size is stored in the semi-dynamic RFID tag, combined with the corresponding fitting segment can obtain semi-dynamic object coverage area. To fit lines and RFID data matching, first we complete RFID tag localization algorithm based on particle filter, and design the location matching and parameter matching algorithm. Finally, introduce the semi-dynamic grid map updating method. Then complete the global topology map update and semantic information database update. Because the map only updates semi-dynamic objects that may move, greatly improve the map updates efficiency.To increase the robot localization means and take full advantage of the location information provided by the static RFID tags distributed in the indoor environment, this thesis proposes robot localization algorithm based on single tag. In order to determine the direction of tag with respect to the robot, we process the antenna probability model with fuzzy inference, get the antenna probability models only related to the angle between the tag and normal of antenna. Then estimate the direction of the tag relative to the robot by Bayesian rules. In order to achieve the more precise robot localization, the artificial landmark based on the laser sensor is designed, and complete the localization algorithm. Finally The algorithm is verified with experiments.