A Collaborative Monocular Visual Simultaneous Localization and Mapping Solution to Generate a Semi-dense 3D Map

Ruwan Janapriya Egodagamage, Purdue University

Abstract

The utilization and generation of indoor maps are critical in accurate indoor tracking. Simultaneous Localization and Mapping (SLAM) is one of the main techniques used for such map generation. In SLAM, an agent generates a map of an unknown environment while approximating its own location in it. The prevalence and affordability of cameras encourage the use of Monocular Visual SLAM, where a camera is the only sensing device for the SLAM process. In modern applications, multiple mobile agents may be involved in the generation of indoor maps, thus requiring a distributed computational framework. Each agent generates its own local map, which can then be combined with those of other agents into a map covering a larger area. In doing so, they cover a given environment faster than a single agent. Furthermore, they can interact with each other in the same environment, making this framework more practical, especially for collaborative applications such as augmented reality. One of the main challenges of collaborative SLAM is identifying overlapping maps, especially when the relative starting positions of the agents are unknown. We propose a system comprised of multiple monocular agents with unknown relative starting positions to generate a semi-dense global map of the environment.

Degree

Ph.D.

Advisors

Tuceryan, Purdue University.

Subject Area

Computer science

COinS