CSC413 – Deep Learning and Neural Networks
Led weekly labs, marked assignments, and ran office hours for ~50 students.
Blerim Abdullai
Hello! I am a masters student at the University of Toronto advised by Florian Skurti and Tim Barfoot where I have been working on localization using radar for surface vessels. I am broadly interested in computer vision with practical applications to robotics and have in the past worked with cameras, lidar, radar and sonar.
Prior to this I completed my bachelors degree at the University of Illinois at Urbana Champaign (UIUC) in computer engineering, where I worked with Julia Hockenmaier on the Alexa Prize Simbot Challenge. I also worked at Pacific Northwest National Laboratory (PNNL) on detecting fish near turbines using passive sonar and autonomous navigation for a surface vessel.
Publications
Hover over some of the publication thumbnails for teaser videos.
RaSCL: Radar to Satellite Crossview Localization
ICRA Workshop on Field Robotics (Spotlight Talk), 2025
Paper / bib
Global localization against RGB satellite images using only mmWave radar on the ground and an initial guess. 50% reduction in error in suburban environments against comparible baselines while maintaining perfomance in urban environments. First paper to localize against RGB images using mmWave radar in marine environments.
PNNL-TUNAMELT: Towards automating the detection of interactions with marine energy devices using acoustic camera sensors
Under Review at Oceanography and Limnology: Methods, 2025
Code
Detecing fish in passive sonar footage pointed at a moving turbine to aid marine biologists in assessing fish interactions near turbines. Used classical methods to reduce video length by 70% while keeping 90% of targets.
BEAST: Building an Embodied Action-prediction System with Trajectory data
Amazon Science, 2023
Paper / bib
Our bot in the 2023 Alexa Prize Simbot Challenge. Featured a text based seq2seq tranformer, MaskRCNN for object detection and grounding, and a dialog managing state machine within a realtime deployment across all Alexa devices. Competition was scored based on user ratings. Made it to semi-finals stage.
Projects
BasedOS
Course Project ECE 391, 2022
x86 Linux kernel built from scratch. 1st place out of 50 teams in a student design competition. Implemented paging, interrupt handling, round robin scheduling with extra features including a buddy allocator, read/write ext2 file system, a network stack up to UDP sockets, GUI with widgets, and multicore support. Parts of this project include completed course assignment materials and cannot be made public, please email me for the code if interested.
Over the Internet Sumobot Outreach
Outreach Live Demos, 2021
Code / Video Led the development of an over-the-internet sumobot project with the College of DuPage Engineering and Technology Club. Features ESP8266 powered robots that can be controlled over the internet via the users arrow keys and a custom WebRTC livestream. This project was used as a fun demo in conjunction with an educational presentation for 5 outreach events.
Teaching
ECE391 – Computer Systems Engineering
Hosted office hours, organized exam review sessions, and graded exams for ~150 students.
ECE385 – Digital Systems Laboratory
Hosted office hours, and review sessions for ~150 students.