Observation of Human Activity in Intelligent Space

点击翻译按钮获取中文摘要

(1) An approach to extract human walking paths independently from the orientation of the paths in a global coordinate system is proposed. More specifically, we propose a similarity measurement based on AMSS(Angular Metrics for Shape Similarity), and then classify human walking paths using a hierarchical clustering method. Experimental results show that the proposed approach achieves rotation invariant extraction of human walking paths.(2) A method to build a map including mobile objects' movement in living environments is proposed. To describe movement history of humans and objects, we detect moving objects from scan data using a laser range finder. The method does not need any background information. Weighting values of each cell on a gird map are calculated based on direction and velocity of movements.(3) The observation system and iSpace are implemented based on Open Robot Technology Middleware(RT-Middleware) to integrate them with a robotics devices embedded environment. This can be used for a platform to build users' activity models and estimate human activity.