Abstract
Mobile robot applications that involve exploration and inspection of dynamic environments benefit, and often even are dependant on reliable novelty detection algorithms. In this paper we compare and discuss the performance and functionality of two different on-line novelty detection algorithms, one based on incremental Principal Component Analysis and the other on a Grow-When-Required artificial neural network. A series of experiments using visual input obtained by a mobile robot interacting with laboratory and real-world environments demonstrate and measure advantages and disadvantages of each approach.
Highlights
Differentiation between common and uncommon stimuli is a desirable ability for mobile robots operating in dynamic environments, since uncommon features often carry the most useful information and deserve to be analysed in more detail
Previous work has demonstrated that the approach of learning models of normality and using them later to highlight abnormalities is very effective for mobile robots that use sonar readings as perceptual input (Marsland et al, 2002)
The GWR network model consists of only 4 vectors while the incremental Principal Component Analysis (PCA) model consists of 35 vectors
Summary
Differentiation between common and uncommon stimuli is a desirable ability for mobile robots operating in dynamic environments, since uncommon features often carry the most useful information and deserve to be analysed in more detail. From an application point of view, reliable novelty detection systems would facilitate the implementation of automated environment exploration and inspection. In these tasks — differently from pattern recognition tasks in which features of interest are already known — one commonly desires to detect any previously unknown entity. The feasible approach to be followed is to learn a model of normality of the environment and use it to filter out abnormal perceptions. Previous work has demonstrated that the approach of learning models of normality and using them later to highlight abnormalities is very effective for mobile robots that use sonar readings as perceptual input (Marsland et al, 2002). This work resulted in the development of the Grow-When-Required (GWR) neural network, a self-organising learning mechanism that is able to classify stimuli as novel or not through the use of a model of habituation
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
