Lexicon acquisition based on object-oriented behavior learning

Studies on lexicon acquisition systems are gaining attention in the search for a natural human–robot interface and a test environment to model the infant lexicon acquisition process. Although various lexicon acquisition systems that ground words to sensory experience have been developed, existing systems have clear limitations on the ability to autonomously associate words to objects. This limitation is due to the fact that categories for words are formed in a passive manner, either by teaching of caregivers or finding similarities in visual features. This paper presents a system for lexicon acquisition through behavior learning. Based on a modified multi-module reinforcement learning system, the robot is able to automatically associate words to objects with various visual features based on similarities in affordances or in functions. The system was implemented on a mobile robot acquiring a lexicon related to different rolling preferences. The experimental results are given and future issues are discussed.