Seminar

Mikkel Baun Kjærgaard is LEGO® Chair and Professor in Software Engineering at the Mærsk Mc-Kinney Moeller Institute of University of Southern Denmark. He holds a Ph.D. degree in Computer Science from 2008. He conducts research within the areas of software engineering and data-driven methods with a specific focus on sensing systems, machine learning and user interaction. His research methodology has an experimental foundation at the intersection of software, machine learning and systems research with applications within Internet of things and Robotics. He has been the PI of many research projects with a high level of industrial involvement. 

Abstract: A challenge for fast robotic automation is that the programming of the automation solutions often requires high level engineering expertise that needs to be built up during extensive university education as well as practical experience with many realized solutions. Furthermore, reliability and efficiency of the user program depend entirely on the knowledge of the engineers in charge. Recent advancements in many fields have been based on the ability to collect, process, and use a wealth of data. This talk will discuss how data can be used to address important problems along the automation journey using data-driven methods. The data-driven methods can be employed to create better environmental models to help robot deployment. They can also be used to create models of robot performance supporting maintenance. Another usage is to help reduce the complexity of robot programming. Potentially, a robot database containing information on previous robot solutions can save time and money for manufacturing companies and enable smaller-scale companies to automate their production as well. Additionally, data-driven methods play a crucial role in AI robotics by enabling robots to learn, adapt, and make informed decisions based on data. The talk will present concrete technology examples from ongoing research and industrial projects (e.g. with Universal Robots). The examples will show recent advancements in data-driven methods for robotics and their industrial application.

Sotaro Fushimi received the Bachelor’s degree of Engineering Science at Kyoto University, Japan in 2025, where he completed his thesis under the supervision of Prof. Toshiyuki Ohtsuka. He is currently pursuing the M.S. degree in the Department of Mechanical and Aerospace Engineering at the University of California, San Diego.

Abstract: In this presentation, we consider a safe controller synthesis of discrete-time stochastic systems using Control Barrier Functions (CBFs). The proposed condition allows the design of a safe controller synthesis that ensures system safety while avoiding the conservative bounds of safe probabilities. In particular, this study focuses on the design of CBFs that provide flexibility in the choice of functions to obtain tighter bounds on the safe probabilities.