Safety and Reliability Despite Uncertainty

Autonomous robotic systems must be able to operate reliably and safely even in the face of uncertainty, disturbances, and incomplete information. Adaptive systems, in particular, pose new challenges for evaluating and verifying their behavior, as their decisions and action strategies can change during operation.

At the Chair of Adaptive Autonomy and Off-Road Robotics, we conduct research into methods for the systematic evaluation, verification, and monitoring of the reliability of autonomous systems. Our engineering approach takes into account both hardware and software throughout the entire system lifecycle.

Research Focuses

Risk Assessment

Autonomous systems must be able to identify potential hazards and assess their impact on people, the environment, and technical systems.

We investigate methods for systematic risk analysis and dynamic risk assessment to detect safety-critical situations and derive risk mitigation measures.

Safety Assurance

The safety of autonomous systems must be substantiated in a transparent manner using appropriate methods, arguments, and evidence.

Our research focuses on safety assurance, safety reasoning, and verification and validation, with particular attention to adaptive and learning system functions.

Evidence-Based Assessment

The reliability of autonomous systems requires robust evidence of their behavior under various operating conditions.

We develop and investigate metrics, test methods, and evaluation procedures to systematically verify safety-relevant properties and identify existing gaps in evidence.

Runtime Assurance

Since not all operational situations are fully predictable, safety-critical properties must also be monitored during operation.

Our focus is on runtime monitoring, dynamic safety assessment, and mechanisms to limit unsafe behavior.

Research Goal

Our goal is to develop reliable autonomous robotic systems whose risks are systematically assessed, whose safety is justifiably demonstrated, and whose behavior can be ensured even under conditions of uncertainty and adaptivity.