Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Towards Semantic SPLAM: Developing a System and Exploring Control Strategies


I. Deutsch

Master Thesis, SS17

We look at the problem of autonomous robot exploration, where in addition to a metric map of the environment, a semantic map of labelled object bounding boxes is created. We refer to this as semantic Simultaneous Planning, Localization, And Mapping (semantic SPLAM). We develop a real-time semantic SLAM base system for trying out different planning strategies, including a semantic mapping system that can process dense point clouds. On top of that, an online exploration goal planner is developed. The planner maximizes the overall expected information gain by continuously taking into account both metric and semantic map information, and guiding the robot to the most promising viewpoint. The functionality of the system is demonstrated in a real system. We further show a statistically significant difference in information gathering efficiency between our planner and two baseline algorithms. To this end, we aggregate data of simulator runs of five minutes each in a variety of environments.

Supervisors: Nikolaos Kariotoglou, John Lygeros, Stefano Soatto


Type of Publication:

(12)Diploma/Master Thesis

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2017:IFA_5689
Permanent link