A PATH PLANNER BASED ON EMOTIONAL INTELLIGENCE FOR TOPOLOGICAL NAVIGATION V. Egido 2 , R. Barber 1 , M. Malfaz, M.A. Salichs 1 1 Carlos III University,Madrid,Spain 2 European University of Madrid, Madrid, Spain veroeg@.uem.es, rbarber@ing.uc3m.es ,mmalfaz@ing.uc3m.es, salichs@ing.uc3m.es Abstract: In this paper a task planning approach based on previous experiences is described. This planner will be used for task planning in a topological navigation system called EDN. The planner has been implemented as one of the deliberative skills of the mobile robot architecture named AD. AD is a two level architecture: deliberative and automatic. In a previous step, the environment’s information is stored by the robot autonomously as a topological map named Navigation Chart. After that, the task planner will be able to get that information for generating different plans. The generated plans are a sequence of actions and events. Two criteria have been established for the plan generation: one that minimizes travel time based on “Dijkstra” algorithm and other that tries to imitate a common human behaviour, minimizing the possible risk using emotions. The last one uses a modified version of Dijkstra algorithm developed for this work called “Guided Dijkstra”. Keywords: Mobile robots, robot skills, planners, robot navigation, emotions. 1. INTRODUCTION As part of everyday life, people navigate from one place to another using their knowledge about the environment. This is a natural process people learn as small children (Ihnelder, et al., 1967) and develop as they grow up. But how do people find their ways? Usually most of the computer models do not simulate the behaviour of human wayfinders. Researches in topological navigation in the last years tend to separate two problems that were before considered as one: path planning and wayfinding. The first one groups together problems related with obstacle avoidance until a defined point is reached. The second one is the task sequencing at higher level. This one can lead in a final event that does not necessarily have to be a specific geometric point. The path planning problem in a first approximation, can be divided in two tendencies. The first one, considered as the classic one, assumes the environment’s total geometric knowledge. The second one, based on the sensorial information, considers the lacks that are linked to the real information obtained of what surrounds such as: partial knowledge, sensor failure, obstacles that do not allow total vision, unexpected situation, etc. Within this field, all the classical avoidance techniques are found, as those based in potential fields or the use of maps based on reticular occupation. With respect to the wayfinding problem, planning movements from an initial place to a final one trying to imitate the human behaviour, researchers in the field of spatial knowledge have the same overall vision of the problem by layers. Timpf and Volta