Signal Processing and Speech Communication Laboratory
hometheses & projects › Position Estimation and Tracking with Uncertain Sensor Positions

Position Estimation and Tracking with Uncertain Sensor Positions

Status
Open
Type
Master Thesis
Announcement date
13 Apr 2021
Mentors
Research Areas

Description: In a smart retail store, the position of each product should be known to guide customers and employees through the store. In general, the positions of the products are unknown and must be estimated since it would be very time consuming to keep track of the positions manually. For that purpose, each product is equipped with a price tag that uses an RF-technology to allow communication and range measurements (received signal strength, time of arrival, …) in between them. The price tags form a wireless sensor network where various state of the art localization algorithms can be applied to estimate their positions. To track a moving object or person through the store, range measurements are performed to price tags and to base stations (for example price tags with exactly known position). The estimated positions as well as the measurements are used to feed a tracking filter. Depending on the used localization algorithm, the estimated position of a price tag could be a point estimate or a posterior distribution which also incorporates the uncertainty of the estimated position. Using the uncertainty of the estimate has a direct impact on the tracking algorithm since it is a measure of how ``trustworthy´´ a measurement to a certain price tag is.

Your Tasks:

  • Review literature/papers on measurement selection and tracking
  • Implement tracking algorithms that deal with uncertainty in the sensor position (e.g. particle filter, particle flow filter, …) for range measurements (RSS, Bluetooth, …) in Matlab
  • Use a performance measure to select only ``trustworthy´´ measurements (e.g. conditional entropy)

Your profile:

  • Master student in information and computer engineering, electrical engineering, audio engineering or similar studies
  • Basic knowledge in statistical signal processing, estimation theory and Bayesian modelling are beneficial
  • Good programming knowledge in Matlab

This thesis is in cooperation with SES-imagotag and the Christian Doppler Laboratory for Location-Aware Electronic Systems at TU Graz. Remuneration of the master’s student by part-time employment is planned.

Contact: Lukas Wielandner (lukas.wielandner@tugraz.at), Erik Leitinger (erik.leitinger@tugraz.at), Klaus Witrisal (witrisal@tugraz.at)