English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Vision is the richest of all human senses: we acquire most of our perceptual information through vision, and perform much of our own vision processing with little conscious effort. In contrast, dealing intelligently with the enormous volume of data that vision produces is one of the biggest challenges to robotics. Identifying an object of interest in a random camera image is a difficult problem, even in domains where the number of possible objects is constrained, such as robotic soccer. This difficulty increases in magnitude when attributes of interest involve change such as movement, and thus require both state information and examining change in visual images over time. Visual analysis also involves many subtle problems, from very lowlevel issues such as identifying colours under changing lighting conditions, to higher-level problems such as tracing the path of an object under conditions of partial occlusion, or distinguishing two objects that are next to one another but appear as one larger object. Humans deal with vision through very specialized massively-parallel hardware, coupled with a broad range of commonsense knowledge. Neither of these is currently feasible to apply to a mobile robot platform. While mobile processors are becoming more powerful, we are still far below what is required to process vision at a human level, and common-sense knowledge has always been one of the most difficult problems associated with artificial intelligence. Vision on mobile robots is thus largely about producing heuristic solutions that are adequate for the problem domain and allow the available hardware to process vision at the frame rate required. Vision in robots may be divided into two types. The first of these, local vision, involves providing a first-person perspective using a camera attached to the robot itself. Local vision shares many of the same problems that humans deal with in visual analysis, such as determining whether motion in a visual stream is due to the motion of objects captured by the camera, or the motion of the agent itself (ego motion). In a local vision setting, each member of a team of robots receives its own unique camera feed and be responsible for analyzing and responding to that feed. There are a number of reasons why local vision may not be preferable in a given application, the foremost of which is the heavy requirement for computational resources. When each robot must perform its own visual processing, it must be able to carry enough on-board processing to do so, which may not be possible in smaller robots or applications with competing resource needs. The fact that each robot is entirely responsible for its own

Intelligent Global Vision for Teams of Mobile Robots