Skip to content. Skip to navigation
CIM Menus
 

Visual Mapping: Representation, Inference and Active Exploration


Dr. Robert Sim
University of British Columbia

January 27, 2005 at  11:30 AM
Conference Room MC603

This work addresses the problem of automatically constructing a visual representation of an unknown environment that is useful for robotic navigation, localization and exploration. There are two main contributions. First, the concept of the visual map is developed, a representation of the visual structure of the environment, and a framework for learning this structure is provided. Second, we examine how a robot can automatically explore an environment and construct a visual map in the presence of odometric uncertainty. This work has important applications in developing visually guided robots for workplace assistance, elder and health care, and hazardous environment operations, such as deep-sea, outer-space and search and rescue domains.


Visual maps model a set of image-domain features extracted from a scene. These features are initially selected using a measure of visual saliency, and subsequently modelled and evaluated for their utility for robot pose estimation. In the first part of this talk we present the details of this framework. Experimental results will be used to demonstrate the feature learning process and the inferred models' reliability for pose inference.


The second part of this work addresses the problem of automatically exploring an environment in order to collect training images and construct a visual map. We demonstrate that a robot's exploration policy impacts the accuracy of its map and examine a variety of approaches to selecting the best policy. These approaches are validated experimentally in both simulated and real-world settings.


The talk will conclude with a discussion of ongoing work in visual representations, and promising future directions for exploring and representing large-scale domains.