Sie sind hier:


English version

ET/IT & Master AR Project Group WS 2012/13

Indoor Localisation and Mapping with Smart Phones



The increasing dissemination of smart phones coincides with their utilization for novel services, such as augmented indoor navigation and guidance within buildings that extend beyond traditional communication and access to information. In GPS based outdoor navigation smart phones replace dedicated navigation systems and become part of everyday life. Similar apps for localization and navigation are hardly available in indoor environments due to the lack of GPS signals and infrastructure for localization. Google just recently started a service that extends Google Maps to selected indoor environments such as airports or malls.  The objective is to replace traditional direction systems, information boards and info points in public places such as airports, train stations or fairs with smart phone technology. Integration of a localization and navigation with smart phone apps enables local services targeted to the facility or organization. The objective of the project group is to develop a prototype system for indoor localization and navigation at selected buildings at TU Dortmund. The following technologies and sensors enable indoor localization

  • Location Fingerprinting (WLAN): The location is inferred by comparison of the signal strength of access points with a radio map that records signal strength at selected reference locations.
  • QR-codes and dead reckoning:  Dead reckoning is a technique from robotics in which the ego-motion is measured by inertial and gyroscopic sensors and compass. The odometry is combined with camera based recognition of locations with unique QR-codes to label locations.
  • Vision based localization: Vision based localization rests upon the comparison of the current view with stored reference views taken at selected locations.

Work package 1 -  Place Recognition and Semantic Mapping

In order to perform complex tasks robots require a deeper spatial and semantic understanding of their environment. In particular for tasks that require collaboration with humans robots depend on a semantic representation of space that they share with humans. The primary goal of the work package is the semantic mapping of an environment by a mobile robot equipped with an omni-directional camera. This includes the classification of places and semantic concepts according to their visual appearance, geometry, topology and presence of objects.

The work package is also concerned with visual place recognition by matching the image captured with the smart phone with a set of reference images taken at previously visited and known locations.

Place recognition and classification rely on local features (SIFT/SURF) as well as global features (CRFH). Training of the classifiers rests upon available datasets (INDECS, IDOL,COLD) as well as novel data acquired in selected buildings and floors at the TU Dortmund. 


    Frank Hoffmann (frank.hoffmann@tu-dortmund.de),

    Felipe Posada (felipe.posada@tu-dortmund.de)

Participants AP1:  2-6 participants



Work package 2 – Mobile Data Collection for Indoor Positioning and Routing

In this work package a data base for the localization and routing techniques should be developed. In order to realize this, a data collection with the help of Android smart phones has to be implemented. With measurements inside the buildings, both local wireless network information (WLAN) and image data have to be collected. The measured data are transmitted directly from the smartphone to a server system and will to be stored in a database. The resulting data pool is used by the other work packages for the development of localization and routing algorithms.


The work package consists of:

  • Implementation of an Android smart phone application for collecting WLAN and image data
  • Setup of a server environment (e.g. JBoss) and a database
  • Realization of a communication link between the smartphone and the server
  • Data measurements in selected buildings of the faculty



    Christian Lewandowski (christian.lewandowski@tu-dortmund.de)

participants A AP2:   3-6 Teilnehmer



Work package 3 – Indoor Positioning with Dead Reckoning and Fingerprinting

Nowadays, smart phones are a part of our daily living. If you want to know where you are, you can use the integrated GPS receiver to localize yourself. Unfortunately, GPS can only be used until you enter a building. Once you enter the building, the localization fails.

This is where the project starts. Instead of using the GPS receiver, other sensors like accelerometers, gyroscopes or WLAN shall be used to enable indoor localization.

Thereby, it is the task of the projectgroup to implement and evaluate applicable indoor localization algorithms which make use of the integrated sensors of a modern smart phone. Furthermore, to enhance the quality of the localization, information of the environment as well as human mobility models shall be considered.


The work package consists of:

  • Investigate applicable localization algorithms and mobility models
  • Implement the localization algorithms in Matlab
  • Implement the mobility models in Matlab
  • Record datasets of typical human movement inside buildings
  • Benchmark and evaluate the algorithms and mobility models
  • Create a demonstrator as client/server architecture for an Android based smart phone



    Daniel Hauschildt (daniel.hauschildt@tu-dortmund.de)

    Theresa Nick (theresa.nick@tu-dortmund.de)

Participants AP3:   3-6 participants




Rudimentary knowledge in Matlab, Java or any other programming language is desirable. Furthermore, a general interest in teamwork, sensors and algorithmic is expected.


No. participants

Alltogether: min 8, max 18


Students in the master program Robotics & Automation are explicitly welcomed to participate in the project group.