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

Novelty detection – the ability to identify perceptions that were never experienced before – and, more generally, selective attention are extremely important mechanisms to autonomous mobile robots with limited computational resources. From the operational point of view, the robot’s resources can be used more efficiently by selecting those aspects of the surroundings which are relevant to the task in hand or uncommon aspects which deserve closer analysis. In fact, identification of new concepts is central to any learning process, especially if knowledge is to be acquired incrementally and without supervision. In order to learn concepts, it is necessary to determine first if they are not already part of the current knowledge base of the agent. Simultaneous learning and recognition (Artac et al., 2002) is a fundamental ability to robots aiming at true autonomy, adaptability to new situations and continuous operation. From the application point of view, reliable novelty detection mechanisms facilitate automated environment inspection and surveillance by highlighting unusual situations. Novelty detection and incremental learning are also vital in applications that demand unsupervised environment exploration and mapping. In this chapter we are particularly interested in environment inspection using mobile robots to assist in the detection of abnormalities. An example of a practical application is the automatic identification of cracks, tree roots or any other kinds of faults in sewer pipes. Sewer inspection is currently performed manually by human operators watching logged video footage, a time-consuming and error-prone approach, due to human fatigue. This task would benefit immensely from the assistance of an inspection robot that is able to pinpoint just the unusual features in the sewer which are likely to correspond to potential faults. Fault detection tasks are different from the usual pattern recognition problems in which the features of interest are usually determined beforehand – the aim in fault detection is to detect something that is unknown a priori. Therefore, it is argued that the most feasible approach to be followed is to learn a model of normality of the environment, and then use it to filter out any abnormal sensory perceptions (Tarassenko et. al., 1995). Abnormal perceptions are thus defined as anything that does not fit the acquired model of normality.

Visual Novelty Detection for Mobile Inspection Robots