3D scene analysis and modeling both for static and dynamic environments

Summary form only given. 3D scene analysis and perception are important tasks for building effective human machine interaction systems. Until now, a large number of techniques have been developed, in computer vision and machine perception fields, for generating 3D descriptions of static and structured environments. However, working environments for modern machines, including robots, have changed into unstructured, dynamic, and outdoor scenes. There emerged big new challenges along with these changes, mainly in perception of both static and moving objects in the scenes. To solve these problems, we carried out researches focused on study of advanced perception systems that can model large-scale static outdoor scenes, track multiple moving targets with frequent interactions, or analyze urban scenes with moving targets. Our research has a number of new features. Different types of 3D sensors are combined to get 3D geometric models of large scenes. Multi-view and multi-type sensors, together with machine learning based algorithms, are used to obtain robust and reliable mapping/tracking results. In addition, a car-based mobile sensor system is developed to explore large sites. This talk presents an overview of the study, in particular focusing on modeling of large scenes, multi-target tracking, and simultaneous 3D mapping and tracking by using the mobile sensing platform. At the same time, it also introduces some case studies about applications of the methods in digital heritage and intelligent transportation systems.