Jon Arrizabalaga
Date:
Speaker
Jon Arrizabalaga is a PhD Candidate at the Technical University of Munich and a visiting researcher at the Robotics Institute at Carnegie Mellon University. Before pursuing his doctoral studies, Jon obtained his MSc. degree at KTH Royal Institute of Technology and wrote the MSc. Thesis at the robotics department of Bosch Research. His research focuses on the convergence of optimization, planning, and control, particularly in their applications to robotics and autonomous systems.
Speaker Links: Website - Google Scholar - X
Abstract
Robotic systems have made remarkable progress in recent years, demonstrating increasingly sophisticated capabilities. However, they remain less reliable, adaptable, and efficient than their biological counterparts. This talk examines a key reason for this gap: the perspective from which actions are decided and executed. While biological organisms act from a first-person, egocentric viewpoint, most autonomous robotic systems rely on third-person abstractions. Motivated by this discrepancy, the talk is based on the premise that adopting an egocentric perspective in control and decision-making can help bridge the performance gap between robots and animals. To this end, we will develop a foundation for egocentricity across all modules of autonomous decision-making—ranging from motion description to environment representation—and demonstrate their application to real-world systems. Central to this framework is a rethinking of the three core elements that define motion: the system’s states, its reference frame, and its perception of the surroundings. By addressing each of these components from an egocentric standpoint, the talk outlines a cohesive and principled framework for building more capable and resilient autonomous systems.
Paper covered during the talk
- Arriza, J., ŠÍR, Z., Manchester, Z., Ryll, M. (2024). A universal formulation for path-parametric planning and control. https://arxiv.org/abs/2410.04664