37 Third Int. Workshop on GNs, Sofia, 1 Oct. 2002, 37-41 A Metamorphic Robot Described in Generalized Nets Terms Zlatogor Minchev Centre for Biomedical Engineering- Bulgarian Academy of Sciences, Acad.G.Bonchev Str., Bl.105, Sofia-1113, BULGARIA e-mail: zminchev@clbme.bas.bg Abstract: The metamorphic robots are a new sphere of robotics that belongs to the Distributed Artificial Intelligence realm. These robots can dynamically adapt their shape and have many useful applications in hostile enviroments on the Earth or on other planets in the space. The metamorphic robot, considered in the paper is described as a multiagent system in generalized nets terms - a new tool for design and simulation that saves the logics and dynamics of the investigated objects but at the same time gives compact and full description. Key words: multiagent systems, generalized nets, modular robots, metamorphic robots, modelling, simulating 1. Introduction The metamorphic robots are a swarm of autonomous controlled units that could connect, disconnect and climb over adjacent modules [1]. As it was discussed in [2], the metamorphic robots could be considered as a multiagent system. The theory of multiagent systems is an extension of artificial intelligence (AI) theory but concerns distributed AI in multiple reasoners [3]. Distributed AI can be devided into two subfields: Distribited Problem Solving (DPS) and Milti-Agent Systems (MAS) [4]. DPS deals with centrally designed systems solving global problems and using build-in cooperations strategies. In contrast, MAS deals with heterogeneous, not necessarily centrally designed agents faced with the goal of a utility - maximizing coexistence [5]. The paper is a combination of both problems - the robots should communicate between themselves in order to solve a common task that is not preliminary known to them. A method for solving such task with a great extent of uncertainty for a chain locomotion and an ameboid movement could be found in [1], [2]. The leading scope of this paper is to present a method for discribing such modules in the terms of generalized nets (GN). The least are a new likelihood that gives an opportunity to describe the dynamics and the logic of a process simultaneously. As the GNs give also a ossibility to describe parallel processes, they could be a convinient tool for ascribing modular robots (metamorphic robots). Each module is ascribed with an identical GN model and there is an element of the generalized net for communication. The revealed model describes the construction of a mechatronic ameboid [1], [2].