Robotic systems are more commonly deployed in open-world environments, where safe and reliable operation is challenged by uncertainty, dynamism, and incomplete knowledge. To explore new approaches for solving those challenges, this work focuses on possibility theory. It explores the use of possibility theory as a framework for reasoning under such conditions. A reasoning solution is proposed, that processes scene graph observations, compares them to existing knowledge, and generates hypotheses using possibility distributions as the foundation for modeling and representing uncertainty. This enables flexible belief updates and action-based inference. The system is evaluated through a set of test scenarios that qualitatively assess its reasoning performance. While still an early-stage implementation, the work highlights the potential of possibility-based reasoning in robotics and lays a foundation for future research in this direction.