Inside the Machine

Abstract

This chapter examines how robots work. All robots can be described in terms of sensors, actuators, overall body parts, and a control system. Control involves generation and coordination of behavior by somehow using incoming sensory information to produce appropriate actuator outputs. Autonomous vehicles are a very good exemplar of current mobile robotics as they employ many of today's cutting-edge techniques. Attempts to develop driverless vehicles go right back to the 1920s. A very significant step in this quest occurred in 1989 when Dean Pomerleau and a team at Carnegie Mellon University (CMU) demonstrated how an artificial neural network could be used to successfully steer a test vehicle around the CMU campus roads. The chapter then considers Pomerleau's system, ALVINN (autonomous land vehicle in a neural network), and studies how the artificial neural networks used in cars and robots work. Current leading-edge autonomous vehicles use a variety of sensors, including vision (cameras), LIDAR, radar, ultrasonics, and GPS information. Finally, the chapter outlines the main approaches to achieving the goal of widespread fully autonomous vehicles, looks at biomechanics, and assesses how reliable and intelligent current robots really are.