Research On Service Robot Holographic Mapping And Its Related Issues Under Home Intelligent Space

With the progress of society and development of robotics, the robots for daily service in families are needed with more urgent. Effective home environment map is the basis for service robots to perceive and adapt to environment, complete path planning, localization and navigation, and ultimately provide highly efficient intelligence services. Different from the industrial environment, home environment of daily life has natural characteristics of diversity, heterogeneity and dynamic. However, subject to the limitations of sensing range and perceive ability, the service robot can not completely build a home environment map reflecting characteristics mentioned above.Intelligent Space technologies effectively compensate for the overall perception of home service robots. The combination of Intelligent Space and robotics can help service robots to sense changes of home environment and objects comprehensively and promptly, so make an effective way for service robots to provide pleasant service in families.In order to reflect the complexities of home environment and meet the needs of robot service mission, the concept of home holographic map is presented. Under the framework of home Intelligent Space and service robots interacting, the key technology of home holographic mapping and its related issues are researched. The main research contents are as follows.Firstly, the current robotic binocular active vision systems and their inadequacies are analyzed, and then a new variable-baseline binocular active vision platform is designed to meet the need of robot service missions. The platform consists of a PTZ camera and a hand-eye camera, so as to obtain a flexible observation capability of multi-direction and multi-resolution. The calibrations of both hand-eye relationship and binocular structural parameters are studied under this platform.Secondly, in order to meet the needs of object detection and tracking in complex background, as well as the needs of robot pleasant operation, an over-complete feature set model of object based on visual features is presented. A method for over-complete feature set extraction is designed utilizing the robot variable-baseline binocular vision, and the method for simultaneous object recognition and pose determination is given based on over-complete feature set. Thirdly, aiming at the perception limitations of robot onboard sensors and the dynamic characteristics of home environment, with the basic idea of Rao-Blackwellized particle filter, an algorithm for simultaneous robot localization and Intelligent Space camera calibration is put forward under the framework of Intelligent Space and robot interacting. The robot pose state is estimated jointing robot control and Intelligent Space camera observations, meanwhile, the camera projection matrix is calibrated jointing robot path and the corresponding Intelligent Space camera observations.Then, for robot to cognitive the dynamic and complex home environment, the concept of home holographic map is proposed integrating information of both environment and its objects. A hierarchical representation model of local geometry and global topology is designed to describe the home holographic map, and the approach to build the holographic map is presented according to both the robot state and object position results with robot binocular vision.Finally, based on the over-complete feature set mentioned above, a particle filter tracking algorithm is given for Intelligent Space camera node to track objects under complex background. The new position of object are calculated based on calibrated Intelligent Space camera and its observation, and then the strategy to update the home holographic map is designed according to the results of tracking and positioning.