Reactive navigation and online SLAM in autonomous frontier-based exploration

Abstract

This paper describes an autonomous exploration algorithm for mobile robots. The method implements a frontier-based exploration strategy that relies on a reactive navigation system and a SLAM algorithm. Despite its strong biological inspiration, the navigation method is specially well suited for the exploration task since it is able to accept and follow higher level position targets while guaranteeing the integrity of the robot. The SLAM module is intended for online execution, but it is able to solve the entire path of the robot in real-time. The hierarchical nature of the SLAM algorithm allows for drift modeling and reduction, which achieves very good resolution maps directly from laser measurements, without extracting landmarks or correction steps. The exploration strategy attempts to exploit the benefits of both the mapping and the navigation algorithms, providing a basic framework for more sophisticated autonomous behaviors. © 2013 Springer-Verlag.