ABSTRACT

Often robots are required to inspect, navigate through, and/or interact with unstructured or dynamically changing environments. Adequate sensing of the environment is the enabling technology required to achieve these tasks. Given advances in technologies such as computational hardware and computer vision algorithms, camera-based vision systems have become a popular and rapidly growing sensor of choice for robots operating in uncertain environments. This chapter focuses on the image geometry obtained between images taken by a single camera at different points in time and space. The geometry obtained from observing four coplanar and non-colinear feature points from a fixed camera is initially described. Euclidean reconstruction of the feature points from image coordinates is then described. The kinematics of a conventional, rigid-link, industrial robot can be conveniently described as a function of joint angles and link lengths using the standard Denavit-Hartenberg convention.