Dissertation defence (Information Technology): MSc Sherif Mohamed
Time
17.2.2023 at 12.15 - 16.15
MSc Sherif Mohamed defends his dissertation in Information Technology entitled “Runtime Resource Management for vision-based Applications in Mobile Robots” at the University of Turku on 17 February 2023 at 12.15pm (University of Turku, Agora, LS XXI, Vesilinnantie 5, Turku).
Opponent: Associate professor Amir Hossein Payberah (KTH Royal Institute of Technology, Sweden)
Custos: Professor Juha Plosila (University of Turku)
Digital copy of the thesis at UTUPub: https://urn.fi/URN:ISBN:978-951-29-9146-4 (copy the link to the browser).
***
Summary of the doctoral dissertation:
This thesis investigates new approaches to improve performance and energy efficiency of computer vision (CV) applications in mobile robots. The goal is to balance accuracy, performance, and energy consumption in real-time. The authors adjust algorithmic and computational parameters and consider the robot's speed as an actuator in their resource management algorithm. Experiments were conducted on an unmanned ground vehicle powered by a Jetson TX2 board and equipped with RGB and event cameras. Results showed noticeable improvement in performance and accuracy while reducing energy consumption compared to state-of-the-art solutions. The authors also found that considering both computational and mechanical aspects together reduced energy consumption more than considering them separately.
Opponent: Associate professor Amir Hossein Payberah (KTH Royal Institute of Technology, Sweden)
Custos: Professor Juha Plosila (University of Turku)
Digital copy of the thesis at UTUPub: https://urn.fi/URN:ISBN:978-951-29-9146-4 (copy the link to the browser).
***
Summary of the doctoral dissertation:
This thesis investigates new approaches to improve performance and energy efficiency of computer vision (CV) applications in mobile robots. The goal is to balance accuracy, performance, and energy consumption in real-time. The authors adjust algorithmic and computational parameters and consider the robot's speed as an actuator in their resource management algorithm. Experiments were conducted on an unmanned ground vehicle powered by a Jetson TX2 board and equipped with RGB and event cameras. Results showed noticeable improvement in performance and accuracy while reducing energy consumption compared to state-of-the-art solutions. The authors also found that considering both computational and mechanical aspects together reduced energy consumption more than considering them separately.
Additional information
University Communications