Pyro Implementation of Swarm-Bots Exploring Target Objects in an Area with Irregular Barriers Byung-Joon Lee, Hyun-Ju Jung, and Gyun Woo Department of Computer Engineering Pusan National University Busan, Republic of Korea {imjuni, hyunjuj, woogyun}@pusan.ac.kr Sung-Youb Jung and Joon-Key Jeon Korea Science Academy Busan, Republic of Korea {thinkgogo, joonkey91}@gmail.com Abstract Robot system is widely used in unmanned exploration or investigation because it can be dangerous or impossible to explore a certain unknown area such as distant planet. For this purpose, highly intelligent single robot systems are generally used. But thinking of disabled situation, a lot of swarm-bots can be alternative to a single highly intelligent robot. This paper reports an implementation of swarm-bots on Player/Stage platform using Pyro. The swarm-bots im- plemented in this paper searching for a set of objects in an area with irregular barriers and bringing them to the home location where they start. A simple map construction algo- rithm is devised and implemented. According to the imple- mentation, the swarm-bots can effectively collect the objects in the exploration area. 1 Introduction Exploring unrevealed area is a challenging task but it can be dangerous or impossible to explore certain areas such as deep-see or distant planets. Intelligent robots take a role of human being in exploring these areas. Sojourner in 1997 or Sprit and Opportunity in 2005 can be typical examples of intelligent robots. However, even if a robot is highly intelligent, it can be- come in a disabled state since the intelligence itself cannot consider every possible situations. Since this kind of ex- ploration is conducted in unmanned or uncontrolled envi- ronment, the exploration can be failed under such disabled situation. Swarm-bots can be an alternative to this situation. Swarm-bots refers a set of robots with swarm intelli- gence. The term ‘swarm intelligence’ was first coined by Beni and Wang [1]. Swarm intelligence can be considered a sum of small intelligence of many agents. The intelligence of an ant colony or the foraging patterns of a group of bees can be an example. Swarm intelligence is experimentally applied in various application areas and some success sto- ries have been reported [8, 9, 7]. However, implementing swarm intelligence is not a sim- ple task. We need a set of robots and the environments where the robots are actuated. Player/State platform devel- oped by Gerkey et al. [5] can be a good experimental envi- ronment for developing swarm intelligence. Pyro [4, 2, 3] on this platform can be a good starting point for program- ming swarm intelligence. This paper reports an implementation of exploring swarm-bots. The goal of the swarm-bots is searching for a set of objects distributed on an area with irregular barri- ers and bringing them together to the home location where the exploration started. Section 2 briefly reviews the swarm intelligence and unmanned exploration. Section 3 and 4 introduces the Pyro programming environment and the Player/Stage platform. Section 5 describes the map con- struction algorithm for the exploration. Implementation re- sults are summarized in Section 6 and Section 7 concludes. 2 Background 2.1 Swarm Intelligence The swarm intelligence, which was firstly coined by Beni and Wang in 1989 [1], is a new approach for problem solving based on collective intelligence of a group of sim- ple entities. And, many technologies based on swarm in- telligence are widely used in various research areas such as computer science, computer network, and robot engineer- ing. We can find the practical case of swarm intelligence application in the social patterns of ants and bees. A sin- gular ant seems to have low-level intelligence, but a large number of ants conduct their affairs intelligently by simple 978-1-4244-2358-3/08/$20.00  2008 IEEE CIT 2008 670