Data-based Risk Analysis and System Optimization

Pozývame vás na prednášku zahraničného profesora Hofbaur na tému Data-based Risk Analysis and System Optimization. Prednáška je otvorená aj pre verejnosť.

Kedy: 5. marca (štvrtok) o 15:00

Kde: prednášková miestnosť CD150

O čom bude prednáška:

State-of-technology systems-design ensures safety of machinery through building upon a thorough risk assessment that subsumes hazard identification, risk estimation and risk evaluation for all operational- and foreseeable unintended-situations. Because of its early stage in a design process, this risk assessment is predominately based on a-priory knowledge and in-depth but conservative expert-analysis and uses a highly structured, standardized procedure to ensure completeness and correctness of this risk modelling procedure. Nevertheless, this probabilistic model builds the basis for the deduction of well-defined, crisp safety requirements for system's design.

This deduction of safety requirements represents a discretization step, where uncertain information of the risk-analysis is translated into specific and well-defined safety constraints (e.g. limiting the speed of operation below a specific maximum, defining timing requirements, etc.). Thus, almost all the system's-design and -revisions are based on the system's safety requirements despite the fact, that the underlying information is subject to uncertain knowledge. Moreover, this uncertain knowledge is mostly based on a-priori information and assumptions.

We propose to introduce a feedback loop that links a system's runtime operation with the associated potential hazards and refining the risk estimation, particularly the probability of hazards, based on facts derived from real-time monitoring of the system's operation to support a better model of risk. We thus introduce a loop-closure, where real-time data obtained through system's monitoring feeds back into risk analysis and risk assessment and thus enabling a forward safety-system refinement in two - standard-conform - steps: First, adapting stringent safety requirements based on the updated risk estimation and secondly, enabling a system's performance revision afore-mentioned safety-system refinement/optimization process.

O hosťovi:

Prof. Hofbaur is Founder & CEO of the deep-tech startup RISKSEN dedicated to building an AI sense of risk for robots in Graz, Austria. In addition, he is Professor for Modular Robotics at the Alpen-Adria University Klagenfurt, Austria, and Visiting Scholar at VSB - Technical University of Ostrava, Czechia. Prior to these roles, he was Director of the Institute of Robotics at the Austrian RTO JOANNEUM RESEARCH and its accredited robot safety testing laboratory, Professor of Automation and Control at the private university UMIT in Hall/Tyrol, Austria, and Visiting Professor/Scholar at M.I.T., Cambridge, USA. He earned his PhD and Habilitation from Graz University of Technology, Austria, in the field of Electrical Engineering specializing on Intelligent Control and Automation of complex Systems in 1999 and 2004, respectively. In addition to his scientific career, he worked as Chartered Engineer specializing on Systems Engineering and Functional Safety. His R&D activities focus on Robotics- and Systems-Safety with foundations in Systems- and Control Theory as well as Artificial Intelligence, dedicated for realizing safe and performant robot systems.

Prednášku organizuje Ústav robotiky a kybernetiky FEI STU