↵ ↵ ↵ ↵ ↵ ↵ ↵ COGNITIVE VISION SYSTEM FOR AN ECOLOGICAL MOBILE ROBOT ↵ ↵ Ana Pavel, Cristian Vasile, Cătălin Buiu ↵ ↵ Faculty of Automatic Control and Computers “POLITEHNICA” University of Bucharest Spl. Independentei 313 Bucharest, Romania email: cbuiu@ics.pub.ro ↵ ↵ ↵ ↵ Abstract: Ecology is a challenging application area for mobile service robots. They should be able to automate tasks that are too tedious or dangerous for humans to execute, such as collecting waste material. Such a robot must make use of several senses, of which the most important and difficult to implement is vision. This paper presents the cognitive vision system of ReMaster One, an autonomous service robot that is able to recognize and sort waste in an indoor environment. A first prototype has been built and tested with success. ↵ Keywords: service robot, ecology, cognitive vision ↵ ↵ ↵ ↵ 1. INTRODUCTION ↵ Over the last 15 years there has been significant progress in the fields of artificial intelligence, computer perception, machine learning and robotics. Yet there has been only minor progress on truly cognitive systems. There are numerous definitions for a “cognitive system”. A recent debate on this topic has attracted more than 40 different answers to “What is a cognitive system?” question (http://www.eucognition.org/wiki/). For example, “cognition is a process of search for an appropriate action by an intelligent agent. An (artificial) cognitive system is one that uses intelligent control, generally modelled on high-level biological intelligent systems; common features are memory, learning, and a capacity for planning” (Joanna Bryson). Or, “cognition is the ability to plan, reason, adapt and act according to high level motivations or goals and using a range of senses, typically including vision, and may be communicate” (Patrick Courtney). Cognition can also be interpreted as “generation of knowledge on the basis of perception, reasoning, learning and prior models” (Christensen, 2003, pp. 17-18). The same Wiki resource lists more than 40 application areas for cognitive systems, ranging from autonomous artificial cells to robot guide dogs. A preferred application area for artificial cognitive systems is robotics. Robots are situated complex systems which are capable to perceive the environment using a wide range of artificial senses and to act upon it after reasoning about their goals, environment, other robots and people. Ecological applications of robotics are generated by a growing interest in keeping the nature clean and the desire to replace humans in tedious, boring, and often dangerous tasks, such as collecting waste. An ecological service robot should be able to search, recognize, pick up and store a significant number of wastes in an indoor or outdoor environment. This paper presents the cognitive vision system of ReMaster One, a prototype of a mobile service robot that is able to recognize, collect and sort empty plastic bottles, empty cans and batteries. ReMaster One has been developed at the Autonomous Robotics Lab of the POLITEHNICA University from Bucharest. Section 2 details the problem of cognitive vision and its relevance to mobile robotics field. Section 3 will briefly present the