Egocentric Decision-Making for Autonomous Systems
Arrizabalaga Aguirregomezcorta, Jon, Ryll, Markus, Technische Universität München · 2025
Robotic systems have advanced rapidly, yet still lag behind animals in adaptability and efficiency. This dissertation explores how adopting an egocentric perspective—acting from the robot's own frame—can bridge this gap. It develops a unified framework for egocentric control and decision-making, including spatial projection of dynamics, local frame motion representation, and smooth safe space modeling. Applications show improved performance across planning, control, and autonomy.