Reasoning about shadows in a mobile robot environment

Abstract

Fenelon, V., Santos, P. E., Dee, H. M., Cozman, F. G. (2013). Reasoning about shadows in a mobile robot environment. Applied Intelligence, 38 (4), 553-565This paper describes a logic-based formalism for qualitative spatial reasoning with cast shadows (Perceptual Qualitative Relations on Shadows, or PQRS) and presents results of a mobile robot qualitative self-localisation experiment using this formalism. Shadow detection was accomplished by mapping the images from the robot?s monocular colour camera into a HSV colour space and then thresholding on the V dimension. We present results of self-localisation using two methods for obtaining the threshold automatically: in one method the images are segmented according to their grey-scale histograms, in the other, the threshold is set according to a prediction about the robot?s location, based upon a qualitative spatial reasoning theory about shadows. This theory-driven threshold search and the qualitative self-localisation procedure are the main contributions of the present research. To the best of our knowledge this is the first work that uses qualitative spatial representations both to perform robot self-localisation and to calibrate a robot?s interpretation of its perceptual input.Peer reviewe

Shadow detection for mobile robots: Features, evaluation, and datasets

“…These measures and the resulting segmentations could form part of a reasoning system which deals with shadows. To date, reasoning systems that have shadows as part of their logic either assume the vision is completed, or use a cut-down environment so that Otsu's method alone is good enough to obtain a shadow segmentation, as noted by Fenelon, Santos, Dee, and Cozman (2013). Although this article does not provide a complete pixels-to-predicates solution for shadow reasoning, we provide a thorough evaluation of the opening stages of such a process and hopefully bring the prospect of fully automated shadow detection and reasoning a little closer.…”

Probabilistic self-localisation on a qualitative map based on occlusions

“…Preliminary results on the framework reported in the present paper were shown in Pereira et al (2013), where a qualitative-probabilistic approach is developed combining the ideas of qualitative localisation using cast shadows proposed in Fenelon et al (2013) with a Bayesian filter. This approach proved to be successful on the tasks of robot localisation and self-calibration of the robots vision system through experiments using a mobile robot in a real environment.…”

“…Following similar ideas, Santos et al (2009) propose a qualitative spatial theory based on a logical formalisation of occlusion and the observation of cast shadows. Preliminary results of applying this theory to a mobile robot domain for the task of localisation were presented in Fenelon et al (2013). Also related to the approach proposed here is the method introduced in McClelland et al…”

“…(Randell et al, 2001)), giving limited application to robotics research. In previous work (Fenelon et al, 2013;Santos, 2007;Santos et al, 2009;Souchanski and Santos, 2008), we have proposed a dynamic formalism about space in which qualitative changes observed by a mobile robot are the building blocks of the system, therefore including the observer in a QSR formalism. These approaches were developed in classical logic languages and, therefore, were not capable of handling sensor uncertainties.…”

Collective singleton-based consistency for qualitative constraint networks: Theory and practice

“…In particular, Bouzy in [24] used RCC8 in programming the Go game, Lattner et al in [25] used RCC5 to set up assistance systems in intelligent vehicles, Heintz et al in [26] used RCC8 in the domain of autonomous unmanned aircraft systems (UAS), and Randell et al in [27] used a particular discrete domain counterpart of RCC8 (called discrete meterotopology) to correct segmentation errors for images of hematoxylin and eosin (H&E)stained human carcinoma cell line cultures. Other typical applications of RCC involve robot navigation [28,29,30], computer vision [31], and natural language processing [32,33].…”