Description :

Machine Learning methods generated some stunning results over the last decades and are increasingly influencing our all lives. Nevertheless there black box character still holds challenges in the the safety assessment. The technique of conformal prediction seems to provide, in theory at least, one mathematical solution for this problem. Nevertheless for ”non-restricted” input this technique can break down. Therefore we need to ask the question, if Conformal Prediction can produce reliable results in the context of Image Input data.

Literature :

Matteo Fontana, Gianluca Zeni, and Simone Vantini. Conformal prediction: a unified review of theory and new challenges, 2022.

more provided after topic assignment

Supervisor: Alexander Günther

Description :

Machine Learning methods generated some stunning results over the last decades and are increasingly influencing our all lives. Therefore the usage of latter ones might be also beneficial in risk assessment for safety critical systems. Nevertheless there black box character still holds challenges in there reliability and trustworthiness. Thus we need to ask the question, if Machine Learning methods can produce reliable results in the context of risk assessment.

Literature :

Machine Learning for Reliability Engineering and Safety Applications: Review of Current Status and Future Opportunities.

More provided after topic assignment

Supervisor: Alexander Günther

Description :

This topic will provide an in-depth exploration of the risk features and their corresponding thresholds in both the situational space and within the Adaptive Cruise Control (ACC) system. The comprehensive examination of this space is critical to understand the myriad of risk features that could potentially affect the functionality of ACC. Defining thresholds for these risk features in the situational space is imperative for discerning deviations and optimizing the response of ACC systems to dynamic highway conditions. A literature survey must be conducted to meticulously define and evaluate thresholds for risk features inherent in ACC systems within highway scenarios, with the aim of developing a profound understanding of the interaction and impact of these features within situational spaces and system components. This research intends to facilitate the enhancement of ACC systems by improving their accuracy and reliability in diverse and dynamic highway environments.

Literature :

(1). Chia, W. M. D., Keoh, S. L., Michala, A. L., & Goh, C. (2021, April). Real-time recursive risk assessment framework for autonomous vehicle operations. In 2021 IEEE 93rd Vehicular Technology Conference (VTC2021-Spring) (pp. 1-7). IEEE.

(2). Fahmy, H. M., Abd El Ghany, M. A., & Baumann, G. (2018). Vehicle risk assessment and control for lane-keeping and collision avoidance at low-speed and high-speed scenarios. IEEE Transactions on Vehicular Technology, 67(6), 4806-4818.

(3). Ammar, H. H., Nikzadeh, T., & Dugan, J. B. (2001). Risk assessment of software-system specifications. IEEE transactions on reliability, 50(2), 171-183.

Supervisor: Anil Patel

Description :

This research endeavor aims to traverse the extensive and varied landscape of risk assessment methodologies, offering a meticulous inquiry into their multidimensional aspects. The exploration focuses on analyzing the principles, frameworks, and applications of diverse risk assessment methodologies. It considers the qualitative and quantitative dimensions, evaluating their relevance and applicability at design time and runtime, and assessing both internal and external risk dimensions. The study also distinguishes between deterministic and data-driven approaches, elucidating their respective merits and demerits. By adopting a comprehensive perspective, the research aims to contribute insightful findings and enhanced understandings to the academic discourse surrounding risk assessment strategies, thereby facilitating more informed and resilient decision-making processes in various domains.

Literature :

(1). Chia, W. M. D., Keoh, S. L., Goh, C., & Johnson, C. (2022). Risk assessment methodologies for autonomous driving: A survey. IEEE transactions on intelligent transportation systems, 23(10), 16923-16939.

(2). Hegde, J., & Rokseth, B. (2020). Applications of machine learning methods for engineering risk assessment–A review. Safety science, 122, 104492.

Supervisor: Anil Patel

Description :

The  growing  interest  in  automated vehicles reveals a gap in traditional safety approaches, which oft en rely on human drivers as a safety measure. The complexity of Automated Driving Systems (ADS) challenges existi ng safety standards like ISO 26262. Industry experts are calling for new safety paradigms and risk models to address these challenges. Advanced methods, such as model-based automated tools, are emerging to assess functi onal safety in this new context. These developments highlight the need for updated safety approaches and regulati ons for autonomous driving. This seminar aims at understanding the functi onal safety and its related concepts. Exploring the open ends and challenges with respect to automated driving systems.

Literature :

• Functional Safety of Automated Driving: Does ISO 26262 meet the challenges?
• Automated Functional Safety Analysis of Automated Driving Systems
• Why functional safety experts worry about automotive systems having increasing autonomy
• Safety goals and functional safety requirements for actuation systems  of automated vehicles
• Enabling functi onal safety ASIL compliance for autonomous driving soft ware systems

Supervisor: Nikita Bhardwaj Haupt

Description :

Ensuring complete safety cases is a key challenge for autonomous vehicles, parti cularly within the Operati onal Design Domain (ODD) where the Automated Driving System (ADS) functi ons. The text off ers four strategies to prevent ADS from exiti ng its ODD, and introduces use cases and a framework to categorize operati ng conditi ons. This modular approach allows for the conti nuous delivery of ADS features. This seminar aims to answer questi ons: What are the most eff ecti ve methods for defi ning and confi ning an ODD  in  ADS? How can use cases be  integrated into  safety models to  ensure  that autonomous  vehicles  operate  safely  within  their  ODD?,  and  what  frameworks  are available or in development for categorizing operati ng conditi ons in the ODD?

Literature :

• Towards an Operational Design Domain That Supports the Safety Argumentati on of an Automated Driving System
• SAE J3016:201806 - SURFACE VEHICLE RECOMMENDED PRACTICE - Taxonomy and Defi niti ons for Terms Related to Driving Automati on Systems for On-Road Motor Vehicles
• Definition and identificati on of system boundaries of highly automated driving
• An  automated  vehicle  safety  concept  based  on  runti me  restricti on  of  the operati onal design domain
• How many operati onal design domains, objects, and events?”

Supervisor: Nikita Bhardwaj Haupt