Research on Robot's Indoor Object Finding Strategy Based on Semantic Relatedness

For the scene of finding things, people will go directly to the area where the purpose of the target object is most closely related, but current robots cannot understand the use relevance of commonly used objects in daily life. In order to give robots an understanding of this content, this article designs an object-finding strategy based on semantic information, which is primarily divided into two parts: visual learning and semantic learning. Visual learning mainly obtains more realistic corpus through the recognition of a large number of actual pictures. Semantic learning mainly processes the corpus and obtains the semantic similarity ranking of objects that are more in line with daily life. Through the verification of the training set and the test set, the semantic similarity ranking obtained in this paper has a good performance.