General Concept of 3D SLAM

Abstract

Simultaneous localization and mapping (SLAM) is a process that fuses sensor observations of features or landmarks with dead-reckoning information over time to estimate the location of the robot in an unknown area and to build a map that includes feature locations. In this chapter, a general model and its related solving algorithm for 3D SLAM are established. The method can be used for all of the situations in the mobile robot community. An underwater mobile robot is used as an example. This chapter is organized as follows: Section 1 is the problem definition; Section 2 establishes all the models for 3D SLAM, including the robot process model, the landmark model, and the measurement model; Section 3 is the method for data association; Section 4 presents the algorithms to solve the SLAM; section 5 describes the multi-sensor related issues based on the underwater mobile robot cases; and Section 6 is the globally-consistent 3D SLAM for mobile robot in real environment.