Probabilistic Integrated Exploration for Mobile Robots in Complex Environments

Abstract

The use of environment maps is one of the main requirements for almost all tasks in mobile robotics. Unfortunately, it is not always possible to have this element, either by the inaccessibility of the environment or simply because there is no usable description of it. The solution to this problem is known as Integrated Exploration or SPLAM (Simultaneous Planning, Localization and Mapping). Considering this problem, we present some strategies based on the extended Kalman filter (EKF), when the mobile robot incrementally builds a map of its environment, while simultaneously using this map for computing the absolute robot localization. At the same time, local decisions on where to move next are performed using the probabilistic strategy-SRT, in order to minimize the error of the estimation of the robot’s pose and the configuration locations. Although the classic strategies have shown good results, there are some inherent problems that prevent achieving the optimal results. For this reason, the paper has focused on creating a SPLAM strategy when the simultaneous localization and mapping is performed using a radical topological approach based on B-spline curves and when the planning of the exploration is conducted using a probabilistic strategy based on graphs.