Rafael Rosales

Sprechstunde:
Sprechzeiten nach E-Mail-Vereinbarung

Rafael Rosales

Prof. Dr.-Ing.

Beruflicher Werdegang

Curriculum Vitae

Prof. Dr.-Ing. Rafael Rosales
Professor of Machine Learning in Production · Lead, Ohm Innovation Factory (OIF)

Professional Experience

06/2026 – 09/2026 · BMW Group, Munich - Principal AI Vehicle Architect

- Project and technical lead for a conversational AI agent for automated driving, with an international team of four developers

- Framework and evaluation pipeline for agent architectures and harnesses; vision-language models for spatial understanding of vehicle surroundings

 

02/2017 – 04/2026 · Intel Labs - Research Scientist

- 2023–2026 · Agentic AI Systems: agentic AI under real-time constraints in collaborative robotics (EU Horizon Europe project MultiX); generative AI for industrial digital twins; agentic framework for scientific code generation (with University of Jena and IIT Delhi); trustworthiness of LLMs in question answering; RL- and LLM-based resource allocation in wireless networks

- 2019–2023 · Dependability and Trustworthy AI Systems: explainable-AI and diversity-based methods for resilient vision and perception models; uncertainty estimation for object detection; knowledge-graph-based safety cases for automated driving (with City, University of London); safety, security and privacy for infrastructure-assisted automated driving

- 2017–2018 · System-Level Modeling, Simulation and Optimization: system-level modeling of V2X baseband SoCs; Python-based simulation and design-space exploration framework, transferred into the commercial tool Intel® CoFluent™ Studio (2021)

- 2017 · Edge-based Collaborative Automated Driving: infrastructure-assisted cooperative perception and quality of information in collaborative driving

- 2011 – 2017 · Friedrich-Alexander-Universität Erlangen-Nürnberg - Research Associate, Chair of Hardware-Software-Co-Design


- System-level modeling of HW/SW systems including power, reliability and temperature; SoC power evaluation; energy-efficient parallel HEVC video coding; time-sensitive networks for aerospace and railway systems

 

Earlier experience
- 2010 Summer· Working Student, BMW Research and Development, Munich
- 2009 Summer· Intern (Extreme Blue), IBM Research and Development, Böblingen
- 2007 – 2008 · Software Engineer and System Validation Engineer, Intel, Guadalajara, Mexico

 

Education


- 2011 – 2016 · Dr.-Ing. (Computer Science), FAU Erlangen-Nürnberg
Dissertation: Holistic Actor-Oriented Modeling of Embedded Systems for ESL Power Consumption Evaluation

- 2008 – 2010 · M.Sc. Communications Engineering, Technische Universität München (TUM)
Thesis: Distributed Augmented Reality Framework, Institute for Human-Machine Communication

- 2002 – 2006 · B.Sc. Comm. and Electronics, Universidad de Guadalajara, Mexico

 

Research Projects

- MultiX (EU Horizon Europe, 2025–2027): research of agentic AI for real-time workloads in industrial manufacturing. 

- 6G-XR (EU Horizon Europe, 2023–2025): wireless communications for XR digital twins.

- Predict-6G (EU Horizon Europe, 2023–2025): Reinforcement learning and generative AI-based resource allocation for deterministic networks.

- ICRI-SAVe (Intel, 2019–2022): safety of autonomous vehicles

- KoRA9 (BMVI, 2017–2019): cooperative radar sensing for the A9 digital test field for infrastructure-assisted automated driving

- KoDak (Bavarian Ministry of Economic Affairs, 2014–2016): time-sensitive networks in aerospace and railway systems

- PowerEval (Bavarian Ministry of Economic Affairs, 2010–2013): power and performance evaluation of mobile radio platforms

 

Awards and Scholarships

- 2021 · Eureka! Award, Intel Labs-wide recognition as top patent submitter

- 2016 · Best Poster Award, PhD Forum, Design, Automation & Test in Europe (DATE)

- 2008 – 2010 · DAAD & CONACyT M.Sc. Scholarship

- 2006 · Mexican Academy of Sciences Summer Research Scholarship, CINVESTAV Guadalajara, Vision and Robotics Group

- 2005 · Advanced Summer School in Physics Scholarship, CINVESTAV Mexico City

 

Lehrgebiete

Machine Learning in industrial production

Forschungsgebiete

Machine Learning in Production

Trustworthy AI

Veröffentlichungen

Publications

Full list: [Google Scholar](https://scholar.google.com/citations?user=FCfJybkAAAAJ)

 

Journal Articles


- K. Komatsu, A. Pauanne, A. Afuwape, S. Hassan, T. Hänninen, **R. Rosales**, J. Kela, T. Rantakokko, E. Piri, J. Prokkola, T. Alves, P. Marques, A. Pouttu, J. Haapola. *Collaborative 3D Digital Twin in Remote Fab Lab over AI-powered Sliced 5G Networks.* EURASIP Journal on Wireless Communications and Networking, 2026. [DOI](https://doi.org/10.1186/s13638-026-02620-x)
- D. Rico-Menendez, P. Picazo-Martínez, C. Barroso-Fernández, A. Calvillo-Fernandez, C. Ayimba, A. de la Oliva, C. J. Bernardos, S. Sudhakaran, **R. Rosales**, D. Cavalcanti. *Experimentation on a Wireless Multi-Domain Deterministic Network: The SLICES-Madrid Approach.* IEEE Communications Magazine, 63(2):54–60, 2025. [DOI](https://doi.org/10.1109/MCOM.001.2400320)
- **R. Rosales**, D. Cavalcanti. *Reinforcement Learning, Rule-Based, or Generative AI: A Comparison of Model-Free Wi-Fi Slicing Approaches.* Frontiers in Signal Processing, 5, 2025. [DOI](https://www.frontiersin.org/journals/signal-processing/articles/10.3389/frsip.2025.1608347)
- A. Bitar, **R. Rosales**, M. Paulitsch. *Gradient-Based Feature-Attribution Explainability Methods for Spiking Neural Networks.* Frontiers in Neuroscience, 17, 2023. [DOI](https://doi.org/10.3389/fnins.2023.1153999)
- **R. Rosales**, M. Paulitsch. *Composable Finite State Machine-Based Modeling for Quality-of-Information-Aware Cyber-Physical Systems.* ACM Transactions on Cyber-Physical Systems, 5(2), 2021. [DOI](https://doi.org/10.1145/3386244)
- **R. Rosales**, M. Glaß, J. Teich, B. Wang, Y. Xu, R. Hasholzner. *MAESTRO – Holistic Actor-Oriented Modeling of Nonfunctional Properties and Firmware Behavior for MPSoCs.* ACM Transactions on Design Automation of Electronic Systems, 19(3), 2014. [DOI](https://doi.org/10.1145/2594481)

 

Conference and Workshop Papers


- **R. Rosales**, D. Cavalcanti. *Position: Cross-Layer Co-Design Is Necessary to Safely Deploy Agentic AI in Real-Time Systems.* IEEE International Conference on Engineering Reliable Autonomous Systems (ERAS), pp. 52–55, 2026. [DOI](https://doi.org/10.1109/ERAS69098.2026.11564461)
- **R. Rosales**, S. Miret. *A Binary Problem in Binary QA: Diverse LLMs or Diverse Question Interpretations? That Is the Ensembling Question.* Language Resources and Evaluation Conference (LREC), 2026. [DOI](https://doi.org/10.63317/43t2yvgid7tw)
- J. Sanson, R. C. Shah, Y. Zhu, **R. Rosales**, V. Frascolla. *WiRD-Gest: Gesture Recognition in the Real World Using Range-Doppler Wi-Fi Sensing on COTS Hardware.* IEEE International Conference on Communications Workshops (ICC Workshops), 2026.
- E. Ferrari, **R. Rosales**, V. Frascolla, D. Cavalcanti. *AI-Enhanced Seamless Handover in Wi-Fi Networks with Multi-Link Management.* International Conference on Software, Telecommunications and Computer Networks (SoftCOM), 2025. [DOI](https://doi.org/10.23919/SoftCOM66362.2025.11197451)
- E. Ferrari, **R. Rosales**, V. Frascolla, D. Cavalcanti. *Multi-Parameter Machine Learning Enhanced Wi-Fi Handover in Dynamic Trajectory Simulations.* IEEE Conference on Standards for Communications and Networking (CSCN), 2025. [DOI](https://doi.org/10.1109/CSCN67557.2025.11230743)
- S. Sudhakaran, J. Perez-Ramirez, D. Cavalcanti, C. Cazan, N. Olson, **R. Rosales**, V. Frascolla. *Wireless Network Digital Twin Calibrated by Real Time Telemetry and XR Feedback Interface.* IEEE International Conference on Factory Communication Systems (WFCS), Toulouse, 2024. [DOI](https://doi.org/10.1109/WFCS60972.2024.10540718)
- **R. Rosales**, P. Munoz, M. Paulitsch. *Assessing the Impact of Diversity on the Resilience of Deep Learning Ensembles: A Comparative Study on Model Architecture, Output, Activation, and Attribution.* IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), Paris, 2023. [DOI](https://doi.org/10.1109/ICCVW60793.2023.00475)
- P. Giardina, P. Szilágyi, C. Chiasserini, J. Carcel, L. Velasco, S. Spadaro, F. Agraz, S. Robitzsch, **R. Rosales**, V. Frascolla, R. Doostnejad, A. Calvillo, G. Bernini. *A Hierarchical AI-Based Control Plane Solution for Multi-Technology Deterministic Networks.* ACM MobiHoc, Washington, 2023. [DOI](https://doi.org/10.1145/3565287.3617605)
- S. Qutub, N. Kose, **R. Rosales**, Y. Peng, M. Paulitsch, K. Hagn, K. Pattabiraman, G. Hinz, A. Knoll. *BEA: Revisiting Anchor-Based Object Detection DNN Using Budding Ensemble Architecture.* British Machine Vision Conference (BMVC), Aberdeen, 2023. [Link](http://proceedings.bmvc2023.org/792/)
- J. Athavale, A. Baldovin, R. Graefe, M. Paulitsch, **R. Rosales**. *AI and Reliability Trends in Safety-Critical Autonomous Systems on Ground and Air.* IEEE/IFIP International Conference on Dependable Systems and Networks Workshops (DSN-W), 2020. [DOI](https://doi.org/10.1109/DSN-W50199.2020.00024)
- **R. Rosales**, G. Wael, F. German. *Actor-Oriented Design Patterns for Performance Modeling of Wireless Communications in Cyber-Physical Systems.* ACM International Symposium on QoS and Security for Wireless and Mobile Networks (Q2SWinet), 2018. [DOI](https://doi.org/10.1145/3267129.3267136)
- **R. Rosales**, C. Herglotz, M. Glaß, A. Kaup, J. Teich. *Analysis and Exploitation of CTU-Level Parallelism in the HEVC Mode Decision Process Using Actor-Based Modeling.* International Conference on Architecture of Computing Systems (ARCS), 2016. [DOI](https://doi.org/10.1007/978-3-319-30695-7_20)
- C. Herglotz, **R. Rosales**, M. Glaß, J. Teich, A. Kaup. *Multi-Objective Design Space Exploration for the Optimization of the HEVC Mode Decision Process.* Picture Coding Symposium (PCS), 2016. [DOI](https://doi.org/10.1109/PCS.2016.7906327)
- B. Wang, Y. Xu, R. Hasholzner, C. Drewes, **R. Rosales**, S. Graf, J. Falk, M. Glaß, J. Teich. *Exploration of Power Domain Partitioning for Application-Specific SoCs in System-Level Design.* MBMV, pp. 102–113, 2016. [DOI](https://doi.org/10.6094/UNIFR/10643)
- **R. Rosales**, M. Glaß, J. Teich. *Mahler: Sketch-Based Model-Driven Virtual Prototyping.* International Conference on Architecture of Computing Systems (ARCS), pp. 85–97, 2014. [DOI](https://doi.org/10.1007/978-3-319-04891-8_8)
- B. Wang, Y. Xu, R. Hasholzner, **R. Rosales**, M. Glaß, J. Teich. *End-to-End Power Estimation for Heterogeneous Cellular LTE SoCs in Early Design Phases.* International Workshop on Power and Timing Modeling, Optimization and Simulation (PATMOS), 2014. [DOI](https://doi.org/10.1109/PATMOS.2014.6951904)
- Y. Xu, B. Wang, R. Hasholzner, **R. Rosales**, J. Teich. *On Robust Task-Accurate Performance Estimation.* Design Automation Conference (DAC), 2013. [DOI](https://doi.org/10.1145/2463209.2488945)
- Y. Xu, B. Wang, **R. Rosales**, R. Hasholzner, J. Teich. *On Confident Task-Accurate Performance Estimation.* International Conference on Architecture of Computing Systems (ARCS), pp. 25–37, 2013. [DOI](https://doi.org/10.1007/978-3-642-36424-2_3)
- S. Glock, F. Reutelhuber, G. Fischer, R. Weigel, T. Ussmueller, **R. Rosales**, M. Glaß, J. Teich. *Scenario-Based Energy Estimation of Heterogeneous Integrated Systems at System Level.* European Microwave Conference (EuMC), pp. 342–345, 2013. [DOI](https://doi.org/10.23919/EuMC.2013.6686661)
- Y. Xu, **R. Rosales**, B. Wang, M. Streubühr, R. Hasholzner, C. Haubelt, J. Teich. *A Very Fast and Quasi-Accurate Power-State-Based System-Level Power Modeling Methodology.* International Conference on Architecture of Computing Systems (ARCS), pp. 37–49, 2012. [DOI](https://doi.org/10.1007/978-3-642-28293-5_4)
- M. Streubühr, **R. Rosales**, R. Hasholzner, C. Haubelt, J. Teich. *ESL Power and Performance Estimation for Heterogeneous MPSoCs Using SystemC.* Forum on Specification and Design Languages (FDL), 2011. [Link](https://ieeexplore.ieee.org/abstract/document/6069489)

 

Dissertation


- **R. Rosales**. *Holistic Actor-Oriented Modeling of Embedded Systems for ESL Power Consumption Evaluation.* Dissertation, Friedrich-Alexander-Universität Erlangen-Nürnberg, 2017. [URN](https://nbn-resolving.org/urn:nbn:de:bvb:29-opus4-81227)

 

Preprints


- **R. Rosales**, P. Popov, M. Paulitsch. *Evaluation of Confidence-Based Ensembling in Deep Learning Image Classification.* arXiv, 2023. [arXiv](https://arxiv.org/abs/2303.03185)

 

Patents

17 granted patents and 18+ pending applications in autonomous driving, human-robot collaboration, perception and digital twins, V2X security and privacy, driver monitoring, and applications of LLMs.

Granted


- *Multimodal Mobility Services with Minimized Perceived Risks.* US 12,460,933 B2, 2025. [Link](https://patents.google.com/patent/US12460933B2/en)
- *Repetitive Task and Contextual Risk Analytics for Human-Robot Collaboration.* US 12,479,101 B2, 2025. [Link](https://patents.google.com/patent/US12479101B2/en)
- *Driver and Environment Monitoring to Predict Human Driving Maneuvers and Reduce Human Driving Errors.* US 12,441,371 B2, 2025. [Link](https://patents.google.com/patent/US12441371B2/en)
- *Generation of Spatial Sound Signal from Auditory Perspective of Individual.* US 12,212,952 B2, 2025. [Link](https://patents.google.com/patent/US12212952B2/en)
- *Systems, Devices, and Methods Involving Driving Systems.* EP 4 015 336 B1, 2025. [Link](https://www.patentguru.com/EP4015336B1)
- *Systems, Methods, and Devices for Reducing Systemic Risks.* US 12,164,367 B2, 2024. [Link](https://patents.google.com/patent/US12164367B2/en)
- *Validation and Training Service for Dynamic Environment Perception Based on Local High Confidence Information.* US 12,082,082 B2, 2024. [Link](https://patents.google.com/patent/US12082082B2/en)
- *Pathloss Drop Trusted Agent Misbehavior Detection.* US 12,063,511 B2, 2024. [Link](https://patents.google.com/patent/US12063511B2/en)
- *Autonomous Driving Vehicle Parking Detection.* TW I802975B, 2023. [Link](https://www.patentguru.com/TWI802975B)
- *System for Modality Selection Monitoring and System for Route Selection Monitoring.* TW I803963B, 2023. [Link](https://www.patentguru.com/TWI803963B)
- *Methods and Devices for Triggering Vehicular Actions Based on Passenger Actions.* US 11,568,655, 2023. [Link](https://patents.google.com/patent/US11568655)
- *Privacy Protection Mechanisms for Connected Vehicles.* US 11,240,659 B2, 2022. [Link](https://patents.google.com/patent/US11240659B2)
- *Controller for an Autonomous Vehicle, and Network Component.* US 11,460,847, 2022. [Link](https://patents.google.com/patent/US11460847B2/en)
- *Securing Vehicle Privacy in a Driving Infrastructure.* US 11,490,249 B2, 2022. [Link](https://patents.google.com/patent/US11490249B2/en)
- *Reconfigurable Network Infrastructure for Collaborative Automated Driving.* US 11,070,988, 2021. [Link](https://patents.google.com/patent/US11070988B2)
- *Broadcasting Map Segments for Individualized Maps.* US 11,003,193, 2021. [Link](https://patents.google.com/patent/US11003193B2)
- *Network Infrastructure for Collaborative Automated Driving.* US 10,403,135, 2019. [Link](https://patents.google.com/patent/US10403135B2)