Machine Learning for Data-Driven Control of Robots

Abstract

Machine learning has become a popular area of research and development in many domains over the past decade. Researchers in a number of disciplines, ranging from biology to economics, have successfully applied machine learning to solve a diverse range of problems including the analysis of protein cell interactions to the prediction of financial market performance. In engineering, machine learning has been used for classification, signal processing, system identification, and control systems. This article provides a discussion of how machine-learning techniques can use data acquired through simulations and experiments to derive more effective sensory abilities, controllers, and decisionmaking strategies for robotic autonomous systems.