Erdal Kayacan

Professor
Paderborn University (Germany)

Brief bio

Erdal Kayacan worked with the School of Mechanical and Aerospace Engineering, Nanyang Technological University (NTU), Singapore, as an assistant professor for four years. He was an associate professor at Aarhus University at the Department of Electrical and Computer Engineering from 2018 to 2023. He is pursuing his career as a full professor at the Department of Electrical and Information Technology at Paderborn University, Germany. His primary research interests and contributions lie within the areas of control systems with applications in guidance, navigation, and automation of unmanned ground, aerial, and underwater vehicles.
 
Dr Kayacan is a member of the Computational Intelligence Society and the Robotics and Automation Society in the Institute of Electrical and Electronics Engineers (IEEE).  He is the IEEE/ASME Transactions Mechatronics technical editor and the associate editor in IEEE Robotics and Automation Letters.
 
 

TALK TITLE

Importance of synthetic datasets for underwater navigation and inspection

 

ABSTRACT

Underwater simulators and datasets push the boundaries of algorithm development for underwater environments. Data availability is particularly challenging for the underwater domain, known for its inaccessibility and the absence of reliable ground truth generation methods. However, a promising solution emerges through using simulators with realistic rendering capabilities capable of generating sensor data and ground truth annotations, effectively addressing data scarcity in underwater scenarios. In the first part of the talk, we will discuss the MIMIR dataset, showcasing a pioneering effort by providing comprehensive ground truth for visual localisation, segmentation, and depth estimation across diverse underwater imaging conditions. In the second half of the talk, we will introduce UNav-Sim. This open-source simulator leverages Unreal Engine 5's high-detail rendering to address the costly underwater robotic surveys' challenges. UNav-Sim offers a realistic virtual environment for testing algorithms and includes an autonomous vision-based navigation stack and compatibility with ROS, streamlining algorithm development. With the ultimate goal of bridging the gap between simulation and reality, UNav-Sim empowers researchers to innovate and test in underwater contexts efficiently​.