Environment Modeling With Automatic Model Selection
技術優勢
No initialization step No restrictions on movement
技術應用
Augmented Reality RoboticsMobile Devices
詳細技術說明
Researchers at the University of California, Santa Barbara have recently created a SLAM system that uses an automatic model selection step to bypass the necessity for any kind of initialization step. This novel SLAM system is capable of automatically choosing an appropriate operation mode based on the inferred camera motion and is able to track the camera and model the environment simultaneously during both rotation-only and parallax-inducing movement. By eliminating the need for an initialization step, the technology is attractive for consumer applications such as mobile device user interfaces.
申請號碼
9495761
其他
Background
One key enabling technology for augmented reality (AR) is the mapping of an unknown environment using sensors while simultaneously localizing the observer in the same environment. This concept is commonly referred to as Simultaneous Localization and Mapping (SLAM). SLAM systems that employ a single camera can only gather depth information if the environment is observed from several distinct vantage points, while using an observer that moves in a non-rotation fashion. Current SLAM systems always require an initialization step and then afterwards camera movement must be restricted to either rotation-only or parallax inducing (non-rotation). SLAM systems that aim at creating a 3D model from a single camera are severely limited by restrictions on movement.
