BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:ce9396ff9975291dba6c38417032acd9
DTSTAMP:20260827T013230Z
SUMMARY:CS MSc Thesis Presentation 31 August 2026
DESCRIPTION:Kontakt: birger.swahn@cs.lth.se\n\nMonday\, 31 August there wil
 l be a master thesis presentation in Computer Science at Lund University\,
  Faculty of Engineering.The presentation will take place in E:4130 (Lucas)
 .Note to potential opponents: Register as an opponent to the presentation 
 of your choice by sending an email to the examiner for that presentation (
 firstname.lastname@cs.lth.se). Do not forget to specify the presentation y
 ou register for! Note that the number of opponents may be limited (often t
 o two)\, so you might be forced to choose another presentation if you regi
 ster too late. Registrations are individual\, just as the oppositions are!
  More instructions for opponents are found here on the LTH thesis project 
 page.15:15-16:00 in E:4130 (Lucas)Presenter: Oscar GrenTitle: Vision-Langu
 age Navigation Models for Mobile RobotsExaminer: Davide TateoSupervisors: 
 Maj Stenmark (LTH)\, Hannes Bergkvist (Levtek)Vision-Language Navigation (
 VLN) models are a class of machine learning models that interpret natural 
 language instructions and vision data to produce navigation instructions f
 or robotics. This thesis compares several different available VLN models a
 nd further fine-tunes a selected model\, DualVLN\, for an Ackermann-contro
 lled industrial mobile robot. Two different real-world navigation tasks we
 re used as target tasks. Results show that fine-tuning the model increased
  performance on the target tasks while it decreased performance on a non-t
 arget task.Furthermore\, this thesis identifies that navigation datasets R
 2R and RxR are fundamentally incompatible with the chosen robot. Analysis 
 shows that only around 1 % of episodes in these datasets can be feasibly e
 xecuted on the robot. To address this\, this thesis introduces an Ackerman
 n-adapted model\, created by finetuning DualVLN on a subset of R2R and RxR
  adapted for the specific Ackermann embodiment. This adapted model outperf
 ormed the base model on two of three real-world tasks.&nbsp\;\n\nMer infor
 mation om händelsen: https://www.cs.lth.se/evenemang/cs-msc-thesis-presen
 tation-31-august-2026
DTSTART;TZID=GMT:20260831T131500
DTEND;TZID=GMT:20260831T140000
LOCATION:E:4130 (Lucas)
END:VEVENT
END:VCALENDAR