arXiv subjects

Thomas Herrmann

Publications and source records attributed to Thomas Herrmann.

11 recordsLinked to original sources

arXiv2026

Intervenability as a Design Requirement for Autonomy and Oversight within Human-Centered AI

Based on the literature and several practical examples of possible AI applica-tions, we outline the concept of intervenability. This new phenomenon is not covered by emergency shutdowns, workarounds, or the reconfiguration of automated systems. Intervenability instantiates the principles of control-lability, autonomy, oversight, and keeping humans in the loop in the context of AI. We provide a taxonomy that encompasses a range of possibilities for intervening activities and differentiates them regarding the mental effort of the users. This taxonomy extends the scope of interventions from real-time control of automated processes to AI-based discrete case-related decision-making. This is in accordance with human-centered AI, which seeks to combine human strengths with the usage of AI. We demonstrate how inter-venability can potentially contribute to the ongoing development of human capabilities on the one hand and to further technical improvement by recon-figuration of AI on the other. Exploring and collaboratively reflecting on the effects of interventions as an integral part of organizational practices is key to enabling this continuous improvement on both sides. Intervenability also provides further momentum for the design of an AI that can help realize in-terventions on its own and advance a smooth transition from intervention to reconfiguration of the AI.

cs.HC↗

arXiv2026

AI as a Sparring Partner -- an HCAI Approach to Promote Human Capabilities

A systematic literature reveals that the role of AI as a sparring partner (SP) is often proposed but not systematically analyzed or defined. We propose the definition: An AI sparring partner (AI SP) interacts with users in a combination of a cooperative as well as a challenging, competitive mode, where AI is selected, customized or self-adapting to meet a level of skills that neither over- nor under-challenges the users. They can have the experience that they become better or are better than AI. AI as a SP can support creativity, extend viewpoints or foster learning and critical thinking either towards ideas and decisions or towards the AI itself. Sparring with AI can either be explicit, e.g. when AI simulates certain roles, or implicit, when AI is used to find out whether or how to perform better.

cs.HC↗

arXiv2026

Organizational Practices and Socio-Technical Design of Human-Centered AI

This contribution explores how the integration of Artificial Intelligence (AI) into organizational practices can be effectively framed through a socio-technical perspective to comply with the requirements of Human-centered AI (HCAI). Instead of viewing AI merely as a technical tool, the analysis emphasizes the importance of embedding AI into communication, collaboration, and decision-making processes within organizations from a human-centered perspective. Ten case-based patterns illustrate how AI support of predictive maintenance can be organized to address quality assurance and continuous improvement and to provide different types of sup-port for HCAI. The analysis shows that AI adoption often requires and enables new forms of organizational learning, where specialists jointly interpret AI output, adapt workflows, and refine rules for system improve-ment. Different dimensions and levels of socio-technical integration of AI are considered to reflect the effort and benefits of keeping the organization in the loop.

cs.HC↗

arXiv2023

Optimal Endurance Race Strategies for a Fully Electric Race Car under Thermal Constraints

This paper presents a bi-level optimization framework to compute the maximum-distance race strategies for a fully electric endurance race car, whilst accounting for the low-level vehicle dynamics and the thermal limitations of the powertrain components. Thereby, the lower level computes the minimum-stint-time for a given charge time and stint length, whilst the upper level leverages that information to jointly optimize the stint length, charge time and number of pit stops, in order to maximize the driven distance in the course of a fixed-time endurance race. Specifically, we first extend a convex lap time optimization framework to capture low-level vehicle dynamics and thermal models, and use it to create a map linking the charge time and stint length to the achievable stint time. Second, we leverage the map to frame the maximum-race-distance problem as a mixed-integer second order conic program that can be efficiently solved in a few seconds to the global optimum with off-the-shelf optimization algorithms. Finally, we showcase our framework for a simulated 6h race around the Zandvoort circuit. Our results show that the optimal race strategy can involve partially charging the battery, and that, compared to the case where the stints are optimized for a fixed number of pit stops, jointly optimizing the stints and number of pit stops can significantly increase the driven distance and hence race performance by several laps.

math.OC↗

arXiv2022

TUM autonomous motorsport: An autonomous racing software for the Indy Autonomous Challenge

For decades, motorsport has been an incubator for innovations in the automotive sector and brought forth systems like disk brakes or rearview mirrors. Autonomous racing series such as Roborace, F1Tenth, or the Indy Autonomous Challenge (IAC) are envisioned as playing a similar role within the autonomous vehicle sector, serving as a proving ground for new technology at the limits of the autonomous systems capabilities. This paper outlines the software stack and approach of the TUM Autonomous Motorsport team for their participation in the Indy Autonomous Challenge, which holds two competitions: A single-vehicle competition on the Indianapolis Motor Speedway and a passing competition at the Las Vegas Motor Speedway. Nine university teams used an identical vehicle platform: A modified Indy Lights chassis equipped with sensors, a computing platform, and actuators. All the teams developed different algorithms for object detection, localization, planning, prediction, and control of the race cars. The team from TUM placed first in Indianapolis and secured second place in Las Vegas. During the final of the passing competition, the TUM team reached speeds and accelerations close to the limit of the vehicle, peaking at around 270 km/h and 28 ms2. This paper will present details of the vehicle hardware platform, the developed algorithms, and the workflow to test and enhance the software applied during the two-year project. We derive deep insights into the autonomous vehicle's behavior at high speed and high acceleration by providing a detailed competition analysis. Based on this, we deduce a list of lessons learned and provide insights on promising areas of future work based on the real-world evaluation of the displayed concepts.

cs.RO↗

arXiv2022

Indy Autonomous Challenge -- Autonomous Race Cars at the Handling Limits

Motorsport has always been an enabler for technological advancement, and the same applies to the autonomous driving industry. The team TUM Auton-omous Motorsports will participate in the Indy Autonomous Challenge in Octo-ber 2021 to benchmark its self-driving software-stack by racing one out of ten autonomous Dallara AV-21 racecars at the Indianapolis Motor Speedway. The first part of this paper explains the reasons for entering an autonomous vehicle race from an academic perspective: It allows focusing on several edge cases en-countered by autonomous vehicles, such as challenging evasion maneuvers and unstructured scenarios. At the same time, it is inherently safe due to the motor-sport related track safety precautions. It is therefore an ideal testing ground for the development of autonomous driving algorithms capable of mastering the most challenging and rare situations. In addition, we provide insight into our soft-ware development workflow and present our Hardware-in-the-Loop simulation setup. It is capable of running simulations of up to eight autonomous vehicles in real time. The second part of the paper gives a high-level overview of the soft-ware architecture and covers our development priorities in building a high-per-formance autonomous racing software: maximum sensor detection range, relia-ble handling of multi-vehicle situations, as well as reliable motion control under uncertainty.

cs.RO↗

arXiv2021

Maximum-distance Race Strategies for a Fully Electric Endurance Race Car

This paper presents a bi-level optimization framework to compute the maximum-distance stint and charging strategies for a fully electric endurance race car. Thereby, the lower level computes the minimum-stint-time Powertrain Operation (PO) for a given battery energy budget and stint length, whilst the upper level leverages that information to jointly optimize the stint length, charge time and number of pit stops, in order to maximize the driven distance in the course of a fixed-time endurance race. Specifically, we first extend a convex lap time optimization framework to capture multiple laps and force-based electric motor models, and use it to create a map linking the charge time and stint length to the achievable stint time. Second, we leverage the map to frame the maximum-race-distance problem as a mixed-integer second order conic program that can be efficiently solved to the global optimum with off-the-shelf optimization algorithms. Finally, we showcase our framework on a 6 h race around the Zandvoort circuit. Our results show that a flat-out strategy can be extremely detrimental, and that, compared to when the stints are optimized for a fixed number of pit stops, jointly optimizing the stints and number of pit stops can increase the driven distance of several laps.

eess.SY↗

arXiv2020

Real-Time Adaptive Velocity Optimization for Autonomous Electric Cars at the Limits of Handling

With the evolution of self-driving cars, autonomous racing series like Roborace and the Indy Autonomous Challenge are rapidly attracting growing attention. Researchers participating in these competitions hope to subsequently transfer their developed functionality to passenger vehicles, in order to improve self-driving technology for reasons of safety, and due to environmental and social benefits. The race track has the advantage of being a safe environment where challenging situations for the algorithms are permanently created. To achieve minimum lap times on the race track, it is important to gather and process information about external influences including, e.g., the position of other cars and the friction potential between the road and the tires. Furthermore, the predicted behavior of the ego-car's propulsion system is crucial for leveraging the available energy as efficiently as possible. In this paper, we therefore present an optimization-based velocity planner, mathematically formulated as a multi-parametric Sequential Quadratic Problem (mpSQP). This planner can handle a spatially and temporally varying friction coefficient, and transfer a race Energy Strategy (ES) to the road. It further handles the velocity-profile-generation task for performance and emergency trajectories in real time on the vehicle's Electronic Control Unit (ECU).

cs.RO↗

arXiv2020

Energy Management Strategy for an Autonomous Electric Racecar using Optimal Control

The automation of passenger vehicles is becoming more and more widespread, leading to full autonomy of cars within the next years. Furthermore, sustainable electric mobility is gaining in importance. As racecars have been a development platform for technology that has later also been transferred to passenger vehicles, a race format for autonomous electric racecars called Roborace has been created. As electric racecars only store a limited amount of energy, an Energy Management Strategy (EMS) is needed to work out the time as well as the minimum energy trajectories for the track. At the same time, the technical limitations and component behavior in the electric powertrain must be taken into account when calculating the race trajectories. In this paper, we present a concept for a special type of EMS. This is based on the Optimal Control Problem (OCP) of generating a time-minimal global trajectory which is solved by the transcription via direct orthogonal collocation to a Nonlinear Programming Problem (NLPP). We extend this minimum lap time problem by adding our ideas for a holistic EMS. This approach proves the fundamental feasibility of the stated ideas, e.g. varying racepaths and velocities due to energy limitations, covered by the EMS. Also, the presented concept forms the basis for future work on meta-models of the powertrain's components that can be fed into the OCP to increase the validity of the control output of the EMS.

eess.SY↗

arXiv2020

Minimum Race-Time Planning-Strategy for an Autonomous Electric Racecar

Increasing attention to autonomous passenger vehicles has also attracted interest in an autonomous racing series. Because of this, platforms such as Roborace and the Indy Autonomous Challenge are currently evolving. Electric racecars face the challenge of a limited amount of stored energy within their batteries. Furthermore, the thermodynamical influence of an all-electric powertrain on the race performance is crucial. Severe damage can occur to the powertrain components when thermally overstressed. In this work we present a race-time minimal control strategy deduced from an Optimal Control Problem (OCP) that is transcribed into a Nonlinear Problem (NLP). Its optimization variables stem from the driving dynamics as well as from a thermodynamical description of the electric powertrain. We deduce the necessary first-order Ordinary Differential Equations (ODE)s and form simplified loss models for the implementation within the numerical optimization. The significant influence of the powertrain behavior on the race strategy is shown.

eess.SY↗

arXiv2016

Your Interests According to Google - A Profile-Centered Analysis for Obfuscation of Online Tracking Profiles

Profiling users for the purpose of targeted advertisements or other kinds of personalization is very popular on the internet. But besides the benefits of individually tailored news feeds and shopping recommendation research has shown that many users consider this practice as privacy infringement. Profiling is often conducted without consent and services offer neither information about what the profile looks like nor which effects it might have. In this paper we argue that understanding profiling and thus interacting with the resulting profiles fosters a privacy literacy that is necessary for users to stay autonomous in the information society. We analyze the extent of interest profiling by google and develop a countermeasures that helps to obfuscate these profiles. To do so we analyzed a links lists from a social bookmarking service with regard to the interests they reveal. We found that, although the profiling by google is very unstable we can still use the information to obfuscate the profile.

cs.CY↗