Pattern Identification of Robotic Environments using Machine Learning Techniques

Abstract

Analysis of time series data collected from mobile robots is getting more attention in many application areas. When multiple robots move through an environment to perform certain actions, an understanding of the environment viewed by each robot is essential. This paper presents analysis of robotic data using machine learning techniques when the data consist of multiple views of the environment. Robotic environments have been classified using the data captured by onboard sensors of mobile robots using a set of machine learning algorithms and their performances have been compared The machine learning model is validated using a test environment where some of the objects are displaced or removed from their designated position.