An Introduction to the Problem of Mapping in Dynamic Environments

Abstract

Robotic mapping comprises one of the most important problems in the field of robotics. During the past two decades, a large number of algorithms have been proposed in order to solve the problem of constructing valid models of the robot environment. As a result, highly accurate maps of large-scale indoor and outdoor environments have been constructed thus far. There are still, though, much to be done in order to achieve fully autonomous mobile robots capable of mapping any kind of environment (structured or unstructured, static or dynamic). In this chapter, we discuss the problem of mapping dynamic environments, an issue that remains open and is extremely active nowadays. Dynamic environments are real world environments where moving objects (e.g. humans, robots, chairs and doors) change their positions over time. Widespread mapping algorithms developed in the past are based on the assumption that the environment remains static during the robot exploration phase. Thus, these algorithms provide imprecise results when applied in non-stationary environments. The need to map these environments has led in the development of new algorithms that are designed to exploit the dynamics of the environments towards efficient mapping. These algorithms have given so far promising results. Through this chapter, we examine the problem of dynamic environments through the mapping point of view. Two issues that are strongly connected to mapping are (a) localization, the process of estimating the position and the orientation of the robot and (b) navigation, the generation of valid paths for the robot. They are both of great importance and remain open fields of research especially when applied on dynamic environments. However, in this chapter we concentrate on the mapping problem and refer to the other two problems only when necessary. Our effort is towards providing the ideas behind the algorithms discussed in this chapter and avoid the mathematical details and formulas. We urge the interested readers to consult the referenced papers in order to gain a better insight on the techniques discussed in this chapter. The outline of the chapter is as follows: In Section 2, we present the mapping problem for static environments, so as to make the reader familiar with the concepts of the mapping problem. Next, in Section 3 we move to the problem of mapping in dynamic environments. More specifically, we discuss the main difficulties of the problem and present a number of