Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
A Study on Long Short-Term Memory Networks Applied to Local Positioning
KTH, School of Industrial Engineering and Management (ITM), Machine Design (Dept.), Mechatronics.
KTH, School of Industrial Engineering and Management (ITM), Machine Design (Dept.), Mechatronics.
2018 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Numerical approaches to lateration and sensor fusion are limited by inherent measure-ment errors and the positioning performance may benefit from alternative approaches. This thesis studies the applicability of deep learning to an Ultra Wide Band (UWB) based local positioning problem in a combination of readings from an Inertial Mea-surement Unit (IMU). Relying on the Robotic Operating System (ROS) and a robotic vacuum cleaner, sensor data was gathered and time stamped in the presence of a ground truth derived from a motion capture system. The gathered time series were processed and then used to train Long Short-Term Memory networks (LSTMs) for predicting two-dimensional coordinates and orientation in a plane. In a series of tests, accuracy and precision of the LSTM predictions were assessed and compared with two conventional approaches to positioning and orientation respectively. Results suggest that LSTMs can be applied well to positioning, however the study failed to establish benefits regarding orientation. It is concluded that the implemented LSTM increased positioning accuracy with 79.9 % and precision with 71.8 % compared to that of the conventional non-linear least mean squares approach. Comparing to the best recorded performance of the LSTM the on-chip sensor fusion with the utilised IMU was 15.8 %more accurate and 70.8 % more precise in estimating orientation from accelerometer, gyroscope and magnetometer readings. Despite this conclusion the study has found results indicating that significant improvements regarding orientating with LSTMs are within close reach.

Abstract [sv]

Numeriska metoder för laterering och sensorfusion är begränsade av inneboende mät-fel och det är möjligt att alternativa tillvägagångssätt skulle gynna prestanda inom positionering. Denna avhandling söker utvärdera frågan genom att undersöka djupin-lärning och dess tillämpning i ett lokalt positioneringssystem baserat på Ultra-Wide-Band (UWB) radio och data från en Inertial Measurement Unit (IMU). Med hjälp av Robotic Operating System (ROS) och en självkörande plattform samlades tidsstäm-plad sensordata in i kombination med den sanna positionen erhållen av ett refer-enssystem. Därefter tränades ett neuronnätverk av typen Long Short-Term Memory (LSTM) till att tillhandahålla uppskattningar av plattformens tvådimensionella ko-ordinater och orientering i planet. Neuronnätverkets prestanda utvärderades sedan i termer av noggrannhet som därefter jämfördes med konventionella tillvägagångssätt för positionering och orientering. Resultaten pekar på att nätverket applicerar väl på positioneringsproblemet, däremot kan inte studien entydigt påvisa förmåga att estimera en orientering. Slutsatserna från denna avhandling är att nätverket ökar noggrannheten för positionering med 79.9 % och precisionen med 71.8 % i jämförelse med ickelinjär minsta kvadratmetod. Jämförelsevis presterar IMU:ns inbyggda sen-sorfusion bättre än nätverket i termer av orienteringsförmåga med 15.8 % och 70.8 %i noggrannhet respektive precision. Trots denna slutsats identifieras resultat som an-tyder att signifikanta förbättringar avseende orientering med LSTM är inom nära räckhåll.

Place, publisher, year, edition, pages
2018. , p. 91
Series
TRITA-ITM-EX 2018 ; 200
National Category
Mechanical Engineering
Identifiers
URN: urn:nbn:se:kth:diva-245183OAI: oai:DiVA.org:kth-245183DiVA, id: diva2:1294130
Supervisors
Examiners
Available from: 2019-03-06 Created: 2019-03-06 Last updated: 2019-03-06Bibliographically approved

Open Access in DiVA

fulltext(13684 kB)85 downloads
File information
File name FULLTEXT01.pdfFile size 13684 kBChecksum SHA-512
c50a378ebb6b671a4a62c2428cc2b6edc30a95013da8644fa8cefcf5606e644a164522a2c0b3c8a39e532235a7e35efb97b9314946ef8e9df5b8a2d0a46d76ea
Type fulltextMimetype application/pdf

By organisation
Mechatronics
Mechanical Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 85 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 231 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf