Machine Learning Components for Autonomous Navigation Systems

Autonomous cyber-physical systems, such as autonomous vehicles, drones etc., are artificial systems that interact with the physical world without any human supervision. Such systems have several intermediate tasks such as sensor data fusion, perception, planning, etc., that are critical for safe operation in the real world. Traditional CPS systems used model-based algorithms for such tasks that required domain knowledge, but in recent years machine learning algorithms have been proposed that are more accurate as they automatically extract pertinent information from large volumes of data. In this chapter, we explore the recent machine learning algorithms that have been proposed to perform the various tasks within an autonomous system.