Blind area measurement with mobile robots Sertan Girgin and Erol S¸ ahin Dept. of Computer Engineering Middle East Technical University ˙ In¨ on¨ u Bulvarı, Ankara, 06531, Turkey E-mail: sertan, erol @ceng.metu.edu.tr Web: http://www.kovan.ceng.metu.edu.tr Abstract. We are interested in the problem of how a mobile robot can measure the area of a closed region that is beyond its immediate sensing range. Our inspiration comes from scout worker ants who assess potential nest cavities. These scouts work literally in dark to assess arbitrary closed spaces. Experimental studies have shown that the scouts can reliably reject nest sites that are small for the colony. These studies support the hypothesis that scouts use the “Buffon’s needle method” to measure the area of the nest. We have implemented the Buffon’s needle method on a simulated mobile robot system and evaluated its performance through systematic experiments. The results show that the method can reliably measure the area of closed regions regardless of their shape and compactness, and that the method is undisturbed by partial barriers placed inside these regions. 1 Introduction Insect studies uncover fascinating examples for autonomous robotics on how, seemingly dif- ficult, problems can be solved through simple behaviors. Based on their observations, biolo- gists make hypotheses about the methods implemented by these simple behaviors. Although these hypotheses are usually tested against experimental data collected from insects, they still beg a constructivist analysis. By constructing behaviors that implement these hypotheses on physically embodied systems, autonomous robotics can not only serve as a testbed for bi- ology, but can also accomplish “technology transfer”, adapting these solutions to artificial systems. In this paper, we study how a mobile robot can measure the area of a closed region that is beyond its immediate sensing. We call this, as the problem of blind area measurement since it resembles to the challenge faced a blind person (lacking a complete view of the region being measured) trying to estimate the area of a large room using his hands (short-range sensing) only. Here, the term blindness denotes that the person (agent) cannot “see” the whole region (either because it is too large or because of objects that occlude a complete view) and that he has to use only his local and short range sensing abilities. There have been quite a number of studies[1, 2, 3] on area coverage of arbitrary closed regions by robots. However, to the best of our knowledge, the question of how the areas of such regions can be measured has not yet been addressed.