Probabilistic Robotics (Intelligent Robotics and Autonomous Agents)
Probabilistic Robotics (Intelligent Robotics and Autonomous Agents)
SURF: speeded up robust features
ECCV'06 Proceedings of the 9th European conference on Computer Vision - Volume Part I
Long-term mapping and localization using feature stability histograms
Robotics and Autonomous Systems
| Hi-index | 0.00 |
Acting in everyday-life environments is still a great challenge in service robotics. Although algorithms and solutions already exist for many relevant subproblems, in particular the aspect of robustness and suitability for everyday use has been neglected so far very often. Robustness and suitability for everyday use are features affecting not only the overall system design but have impact on each single algorithm of each component. Although an overwhelming amount of work is available to address the SLAM problem, the challenge of applying a SLAM algorithm over the whole lifecycle of a service robot, perhaps even in different environments, has not been brought into focus very often. An obvious problem to be solved is the continuously growing number of landmarks. A lifelong running SLAM approach requires means to select landmarks such that they best cover the working environment given bounded SLAM resources like the maximum number of manageable landmarks. This paper proposes a novel solution for selecting appropriate landmarks to limit the number of landmarks. The idea is to quantify the contribution of a landmark to the ability of the robot to localize itself in its working environment. Thus, the core contribution is to base the landmark selection process upon the landmarks' coverage of the working environment. Real-world experiments on a P3DX-platform with a bearing-only SLAM approach and an omnicam confirm that the addressed question and the proposed first approach might be another step towards the overall goal of suitability for everyday use.