Learning organizational principles in human environments

摘要: In the context of robotic assistants in human everyday environments, pick and place tasks are beginning to be competently solved at technical level. The question where objects or them up from, among other higher-level reasoning tasks, is therefore gaining practical relevance. this work, we consider problem identifying organizational structure within an environment, i.e. determining principles that would allow a robot infer best particular, previously unseen object reasonably search for particular type given past observations about allocation locations environment. This can formulated as classification task. We claim governed by notion similarity provide empirical analysis importance various features datasets describing kitchens. For aforementioned compare standard methods, reaching average accuracies least 79% all scenarios. thereby show that, ontology-based measures well-suited highly discriminative features. demonstrate use learned models kitchen environment on real system, identifies newly acquired item, determines suitable location then stores item accordingly.