Design and analysis of visualization techniques for mobile robotics development

Abstract

Simultaneous localization and mapping (SLAM) algorithms are of vital importance in mobile robotics. This paper presents novel Augmented Reality (AR) visualization techniques for SLAM algorithms, with the purpose of assisting algorithm development. We identify important algorithm invariants and parameters and combine research in uncertainty visualization and AR, to develop novel AR visualizations, which offer an effective perceptual and cognitive overlap for the observation of SLAM systems. A usability evaluation compares the new techniques with the state-of-the-art inferred from the SLAM literature. Results indicate that the novel correlation and color-mapping visualization techniques are preferred by users and more effective for algorithm observation. Furthermore the AR view is preferred over the non-AR view, while being at least similarly effective. Since the visualizations are based on general algorithm properties, the results can be transferred to other applications using the same class of algorithms, such as Particle Filters.