Output details
11 - Computer Science and Informatics
University of Edinburgh
Temporally scalable visual SLAM using a reduced pose graph
<02> Originality: Outlines a complete system for Visual SLAM on a much larger scale than previously reported.
Significance: Illustrates that visually-based approaches to mobile robotics within natural environments are practical and closer to maturity. Demonstrated on a large scale (50km long), which exposed and proposed consistent solutions to problems of scale: principled reduction of large-scale pose graphs and the demonstration of the leading computational constraints of the problem. Nominated (4 of 2000 submissions) for ICRA 2013 best student paper.
Rigour: Extensive ground-truth evaluation of the experimental results to provide a benchmark for future comparison (published separately by IJRR).