Building, Understanding, and Reflecting on Knowledge

Autonomous robotic systems must not only perceive their environment, but also be able to gain knowledge from observations, develop that knowledge over time, and assess its reliability. Especially in dynamic and incompletely observable environments, how robots interpret information, store experiences, and question existing assumptions is crucial.

At the Chair of Adaptive Autonomy and Off-Road Robotics, we investigate methods that enable autonomous systems to build knowledge, gain insights, and reflect on their own knowledge bases. In doing so, we examine the interplay between perception, memory, knowledge representation, and reasoning.

Research Focuses

Perception and Understanding

Sensor data initially provides individual observations from which autonomous systems must develop an understanding of their environment.

We investigate methods for interpreting and semantically abstracting perceptual information in order to identify relevant properties, relationships, and changes in the environment.

Knowledge Construction and Memory

Autonomous systems require mechanisms to store information beyond individual observations and make it available for future situations.

Our research focuses on short-term and long-term memory, knowledge representation, and the continuous updating of environmental models. The central question is which information remains relevant in the long term.

Knowledge Acquisition

Knowledge is not created solely by collecting data, but also by linking, interpreting, and evaluating it.

We investigate how autonomous systems can derive new insights from observations and experiences, develop expectations about their environment, and test these expectations against new information.

Knowledge Reflection

Autonomous systems must be able to assess how reliable their knowledge is, what information is missing, and when existing assumptions should be reviewed.

The focus is on evaluating the quality of knowledge, identifying contradictions, and revising existing beliefs based on new observations.

Research Goal

Our goal is to develop knowledge-capable autonomous robotic systems that not only process information but can also build knowledge, utilize experience, gain new insights, and take into account the limits of their own knowledge.