Reliable Robot Localization

6A Reliable Temporal Approach for the SLAM Problem

6.1. Introduction

6.1.1. Motivations

Research on SLAM topics is relatively recent (Smith et al. 1990) and gathers today a wide part of the robotic community. Several research issues have triggered the development of numerous SLAM solutions, mainly based on probabilistic approaches. The literature on this topic is wide and several books and papers have been written to capture the full picture of the various investigated methods (see, for example, (Durrant-Whyte and Bailey 2006, Bailey and Durrant-Whyte 2006, Thrun and Leonard 2008)).

In SLAM, the noise coming from measurements is usually handled in a probabilistic way, while less attention has been paid to set-membership solutions (Yu et al. 2016, Di Marco et al. 2001, Jaulin 2011). The latter have the advantage of providing a reliable quality assessment of the localization and mapping, which might be expected for safety or military applications. In the underwater case, for example, a public service in hydrography and maritime cartography will have to elaborate accurate maps for navigation purposes. We can easily understand the issues related to inaccurate maps and thus the need to comply with standards for hydrographic surveys (see, for example, Table 6.1).

AUVs involved in surveys will have to precisely estimate these uncertainties. In addition, for practical reasons mentioned in the introduction of this book, AUVs might have to survey without surfacing, and thus a SLAM ...