A CNN Regression Approach to Mobile Robot Localization Using Omnidirectional Images

Understanding the environment is an essential ability for robots to be autonomous. In this sense, Convolutional Neural Networks (CNNs) can provide holistic descriptors of a scene. These have proved robust in dynamic environments. The aim paper perform hierarchical localization mobile robot indoor by means CNN. Omnidirectional images are used as input Experiments include classification study which CNN trained so that able find out room where it located. Additionally, transfer learning technique transforms original into regression estimate coordinates position specific room. Regarding classification, retrieval task performed with considerable success. As stage, when along approach based on splitting rooms, also provides relatively accurate results.