Multi-Services Collaborative Optimization Based On Improved Data Fusion Tree In Intelligent Space

With the development of computer and communications technology, home service robot with the capability of human-robot coexistence have gradually stepped into our life. Currently, the information (such as environmental information and human behavior) that robot obtained with its onboard sensors is limited, and the accuracy is depended on the performance of sensors. For service robot, how to percept the environmental change, understand human’s behavior and provide autonomous, accurate and smart service in complicated home environment are important problems. To solve the above problems, the paper designs the intelligent space and cloud database, it can obtain more comprehensive environmental information, perceive and track human more completely based on information fusion in intelligent space.Firstly, according to the multi-source heterogeneous property of information under the complex scenarios, we establish the available and inherited data fusion tree of multi-service optimizing collaborations. Based on the data features and service types, the information pretreatment technology of multi-sensor is studied. The space-temporal calibration and data fusion algorithm is designed to realize the multi-level fusion model.Secondly, in the distributed data fusion tree, the paper studies information acquisition, preprocessing, feature representation and tracking method of the wireless sensing devices and distributed vision. Based on multi-source heterogeneous scene and posture information, it proposes the target tracking model of multi-level, multi-sensor and multi-algorithm to solve the target missing and other issues, and it can obtain more accurate three-dimensional target information and tracking effect.Thirdly, based on the target coordinates of global environment, robot make up the problem of the relatively narrow perception using information fusion. Firstly, when the moving target is detected, the system will scan target human’s leg by laser range finder to cluster the nearest neighbor, and the exact distance information will be got. Secondly, this paper adopts a better Extend Kalman Filter for heterogeneous sensor information fusion to realize the simultaneous robot localization and target tracking in intelligent space. At last, the experimental results verify the effectiveness of the proposed method.Fourthly, based on network service mechanism, these modules mentioned in above chapter are integrated and a server framework is built to construct an intelligent software and cloud database system in the intelligent space. The real operation of each module according to the daily service needs and emergencies are verified.