Lecturer (Assistant Professor) in Autonomous Systems · Queen's University Belfast

Hakan Yekta Yatbaz

I build introspective perception for autonomous systems that know when they are failing.

I am a Lecturer (Assistant Professor) in Autonomous Systems in the School of EEECS at Queen's University Belfast. My research focuses on making AI in autonomous and intelligent systems safe, reliable, and trustworthy across robust sensing, run-time monitoring and introspection, explainable representations, and cooperative autonomy. Before joining Queen's, I was a Postdoctoral Research Fellow at WMG, University of Warwick (working on the EU-funded EVENTS and Hi-Drive projects), where I also completed my PhD advised by Dr. Roger Woodman and Prof. Mehrdad Dianati.

PhD opportunities Selected work

Recruiting Funded PhD studentships (DfE Northern Ireland & UK/Irish Common Travel Area Home fees)

01 - News

Recent

Oct 2026
Gave an invited seminar for the IEEE QUB Student Branch: "Can We Trust AI in Autonomous Systems? From Perception to Run-Time Error Detection".
2026
Appointed Associate Editor for Connected Mobility and Automation, Frontiers in Future Transportation.
Sept 2025
Joined Queen's University Belfast as Lecturer in Autonomous Systems (School of EEECS).
Feb–Sept 2025
Postdoctoral Research Fellow at WMG, University of Warwick, working on the EU-funded EVENTS and Hi-Drive projects.
Oct 2024
Submitted PhD thesis on run-time error detection for DNN-based perception in automated driving systems.
Oct 2025
Multi-Layer Self-Assessment with Filtering for 3D Object Detection published in ACM Transactions on Intelligent Systems and Technology (TIST).

Earlier news

Sept 2024
Integrity Monitoring of 3D Object Detection using Spatial Filtering presented at IEEE ITSC 2024 (Edmonton, Canada).
Jun 2024
Early-Layer Neural Activation Monitoring for 3D Detection presented at IEEE/CVF CVPRW 2024 (SAIAD Workshop).
May 2024
Run-Time Introspection of 2D Object Detection Using Learning Representations published in IEEE Transactions on Intelligent Vehicles (T-IV).
Feb 2024
Survey on Introspection of DNN-Based Perception in Automated Driving published in IEEE Transactions on Intelligent Transportation Systems (T-ITS).

02 - Throughline

Sensing under noise, introspection at run time, representations that explain, autonomy that cooperates.

Each dot is a paper: from multimodal anomaly sensing to neural activation monitoring, and next, resilient cooperative autonomy.

2019 Scanpath STA IEEE Sensors 2019 · Sensing
2020 Signature Stress IEEE IWBF 2020 · Xai
2021 Color-Coded CNN IEEE Sensors 2021 · Xai
2022 Hazardous Events IEEE ITSC 2022 · Autonomy
2022 Forest Fire WMSN ACM TOMM 2022 · Sensing
2023 E-Health IoT IoT Elsevier 2023 · Sensing
2023 ICCVW Patterns ICCVW 2023 · Introspection
2024 T-ITS Survey IEEE T-ITS 2024 · Introspection
2024 T-IV Introspection IEEE T-IV 2024 · Xai
2024 Early-Layer 3D CVPRW 2024 · Xai
2024 Spatial Filtering IEEE ITSC 2024 · Autonomy
2025 Audio-IMU HAR Sci. Rep. 2025 · Sensing
2025 Multi-Layer 3D ACM TIST 2025 · Introspection
2026 Cooperative Autonomy Direction · Autonomy
Next Your PhD? See open problems

Hover, tap or tab to a dot. Accent dots mark flagship works.

Sensing

Multi-modal 2D/3D perception under weather and sensor noise.
Introspection

Run-time monitors that flag perception errors without ground truth.
Representations

Early-layer neural activations and compact latent encodings.
Autonomy

Spatial filtering and collaborative V2X for resilient autonomy.

03 - Experience

Education and experience

Education

Middle East Technical UniversityB.Sc. & M.Sc. in Computer Engineering2013–2020
University of Warwick (WMG)Ph.D. in Engineering (Autonomous Systems)2020–2024

Work

METU NCCGraduate Teaching Assistant2018–2020
University of WarwickGraduate Teaching Assistant2020–2025
WMG, University of WarwickPostdoctoral Research FellowFeb–Sept 2025
Queen's University BelfastLecturer (Assistant Professor) in Autonomous SystemsSept 2025–now

04 - Selected work

Three flagship works, opened up.

Architectures, key findings, and code for run-time perception monitoring and self-assessment in automated driving.

Plate IArchitecture schematic

Multi-layer activation extraction across early, middle, and deep 3D backbone stages combined with spatial ego-corridor filtering to flag silent 3D detection failures at run time. ACM TIST 2025.

Plate I · ACM TIST 2025 · 3D LiDAR Introspection

Multi-Layer Self-Assessment

Filtering and cross-layer activation monitoring for 3D object detection

Deep 3D object detectors can fail silently when encountering adverse weather, sparse point clouds, or out-of-distribution road scenes. Instead of relying on a single late feature map or miscalibrated confidence scores, this framework extracts activation signatures across multiple backbone depths and applies spatial ego-corridor filtering so the safety monitor focuses on hazards that matter to the vehicle's trajectory.

Backbone feature hierarchy: MultiLayer
Safety-critical ROI focus: SpatialFilter
Published in: ACMTIST '25

See publication entry Paper (DOI) Code

Plate IITaxonomy schematic

Unified taxonomy of DNN perception introspection in automated driving, contrasting input reconstruction, internal activation monitoring, and output uncertainty estimation. IEEE T-ITS 2024.

Plate II · IEEE T-ITS 2024 · Survey & Taxonomy

Introspection of DNN Perception

State-of-the-art and open research challenges in automated driving

How can an automated vehicle tell that its deep neural perception pipeline is making a critical mistake without ground-truth labels at run time? This work synthesises the field of DNN introspection across 2D/3D detection and segmentation, formalising failure triggers, evaluation metrics, and the latency–reliability trade-offs required for safety-critical deployment.

Google Scholar citations: 36+Citations
IEEE ITS Society flagship: T-ITS2024

See publication entry Paper (IEEE Xplore)

Plate IIIMethod schematic

Learning compact representations from early and intermediate neural activation patterns enables faster, more reliable error detection than waiting for final detector layers. IEEE T-IV 2024 & CVPRW 2024.

Plate III · IEEE T-IV 2024 & CVPRW 2024 · Activation Representations

Early-Layer & Representation Introspection

Detecting 2D and 3D perception errors before final-layer suppression

Prior introspection methods almost exclusively inspect the penultimate layer of a detector—where features of missed objects have already been attenuated. Across our IEEE T-IV and CVPRW SAIAD papers, we demonstrated that early-layer neural activations and learned representation embeddings capture rich failure cues while cutting monitoring latency.

2D & 3D object detection: 2D/3DVision & LiDAR
Activation tap depth: EarlyBackbone

See publication entries IEEE T-IV CVPRW PDF arXiv

Code & open research

Open source

multi-layer-introspection

Reference implementation for multi-layer self-assessment and spatial filtering in 3D LiDAR object detection (ACM TIST 2025).

github.com/yatbazhakan/multi-layer-introspection
Preprints & papers

Google Scholar profile

Full citation record, open-access preprints, and bibliometric indices across autonomous driving and intelligent sensing.

scholar.google.com
Repositories

GitHub (@yatbazhakan)

Codebases, experimental scripts, and reproducibility packages for DNN introspection and sensor analytics.

github.com/yatbazhakan

05 - Publications

Publications

Showing 20 of 20 publications

2026

MMM 2026 ConferenceAI Platforms / NLP

ZhadigerAI: Software-as-a-Service AI Platform for Kazakh and English

A. Baglanova, N. Fakhrutdinov, U. Azirakhmet, R. Kalimzhanov, Z. Babakhojayeva, H. Y. Yatbaz, A. Yazıcı

32nd International Conference on Multimedia Modeling (MMM 2026) · Lecture Notes in Computer Science (LNCS), Springer Nature

A bilingual Software-as-a-Service AI platform integrating speech and language processing pipelines for Kazakh and English.

DOI

BibTeX

@inproceedings{baglanova2026zhadigerai,
  title     = {{ZhadigerAI}: Software-as-a-Service {AI} Platform for {Kazakh} and {English}},
  author    = {Baglanova, A. and Fakhrutdinov, N. and Azirakhmet, U. and Kalimzhanov, R. and Babakhojayeva, Z. and Yatbaz, H. Y. and Yaz{\i}c{\i}, A.},
  booktitle = {International Conference on Multimedia Modeling (MMM)},
  series    = {Lecture Notes in Computer Science},
  publisher = {Springer Nature},
  year      = {2026},
  doi       = {10.1007/978-981-95-6963-2_11}
}

2025

ACM TIST 2025 JournalAutonomous Systems

Multi-Layer Self-Assessment with Filtering for 3D Object Detection in Autonomous Vehicles

H. Y. Yatbaz, K. Koufos, M. Dianati, R. Woodman

ACM Transactions on Intelligent Systems and Technology (TIST), Vol. 17, No. 2

Fuses neural activations across multiple backbone layers with spatial filtering to detect 3D LiDAR perception errors at run time.

DOI Code

BibTeX

@article{yatbaz2025multilayer,
  title   = {Multi-Layer Self-Assessment with Filtering for 3D Object Detection in Autonomous Vehicles},
  author  = {Yatbaz, Hakan Yekta and Koufos, Konstantinos and Dianati, Mehrdad and Woodman, Roger},
  journal = {ACM Transactions on Intelligent Systems and Technology (TIST)},
  volume  = {17},
  number  = {2},
  year    = {2025},
  doi     = {10.1145/3776551}
}

Sci. Rep. 2025 JournalMultimodal Sensing

Hierarchical Human Activity Recognition with Fusion of Audio and Multiple Inertial Sensor Modalities

T. A. Yılmaz, H. Y. Yatbaz, E. Ever, A. Yazıcı

Scientific Reports (Nature Portfolio), 15(1)

A hierarchical multi-modal fusion architecture combining acoustic cues and distributed inertial sensors for robust human activity recognition.

DOI

BibTeX

@article{yilmaz2025hierarchical,
  title   = {Hierarchical Human Activity Recognition with Fusion of Audio and Multiple Inertial Sensor Modalities},
  author  = {Y{\i}lmaz, T. A. and Yatbaz, Hakan Yekta and Ever, Enver and Yaz{\i}c{\i}, Adnan},
  journal = {Scientific Reports},
  volume  = {15},
  number  = {1},
  year    = {2025},
  doi     = {10.1038/s41598-025-29801-w}
}

2024

IEEE ITSC 2024 ConferenceAutonomous Driving

Integrity Monitoring of 3D Object Detection in Automated Driving Systems using Raw Activation Patterns and Spatial Filtering

H. Y. Yatbaz, M. Dianati, K. Koufos, R. Woodman

27th IEEE International Conference on Intelligent Transportation Systems (ITSC), Edmonton, Canada

Combines raw internal neural activations with ego-vehicle spatial context filtering to focus integrity monitoring on safety-critical zones.

DOI arXiv PDF

BibTeX

@inproceedings{yatbaz2024integrity,
  title     = {Integrity Monitoring of 3D Object Detection in Automated Driving Systems using Raw Activation Patterns and Spatial Filtering},
  author    = {Yatbaz, Hakan Yekta and Dianati, Mehrdad and Koufos, Konstantinos and Woodman, Roger},
  booktitle = {27th IEEE International Conference on Intelligent Transportation Systems (ITSC)},
  year      = {2024},
  doi       = {10.1109/ITSC58415.2024.10920125}
}

CVPRW 2024 ConferenceAutonomous Driving

Run-Time Monitoring of 3D Object Detection in Automated Driving Systems Using Early Layer Neural Activation Patterns

H. Y. Yatbaz, M. Dianati, K. Koufos, R. Woodman

IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), SAIAD Workshop

Shows that early-layer LiDAR backbone activations carry strong failure signatures before errors propagate to 3D bounding-box heads.

DOI arXiv PDF (CVF)

BibTeX

@inproceedings{yatbaz2024earlylayer,
  title     = {Run-Time Monitoring of 3D Object Detection in Automated Driving Systems Using Early Layer Neural Activation Patterns},
  author    = {Yatbaz, Hakan Yekta and Dianati, Mehrdad and Koufos, Konstantinos and Woodman, Roger},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)},
  pages     = {3518--3527},
  year      = {2024},
  doi       = {10.1109/CVPRW63382.2024.00356}
}

IEEE T-IV 2024 JournalAutonomous Driving

Run-Time Introspection of 2D Object Detection in Automated Driving Systems Using Learning Representations

H. Y. Yatbaz, M. Dianati, K. Koufos, R. Woodman

IEEE Transactions on Intelligent Vehicles (T-IV), Vol. 9, No. 5, pp. 4880–4892

Learns compact, interpretable representations of detector feature maps to flag silent false negatives in 2D automated driving perception.

DOI arXiv PDF

BibTeX

@article{yatbaz2024tiv,
  title   = {Run-Time Introspection of 2D Object Detection in Automated Driving Systems Using Learning Representations},
  author  = {Yatbaz, Hakan Yekta and Dianati, Mehrdad and Koufos, Konstantinos and Woodman, Roger},
  journal = {IEEE Transactions on Intelligent Vehicles},
  volume  = {9},
  number  = {5},
  pages   = {4880--4892},
  year    = {2024},
  doi     = {10.1109/TIV.2024.3385531}
}

IEEE T-ITS 2024 JournalAutonomous Driving

Introspection of DNN-Based Perception Functions in Automated Driving Systems: State-of-the-Art and Open Research Challenges

H. Y. Yatbaz, M. Dianati, R. Woodman

IEEE Transactions on Intelligent Transportation Systems (T-ITS), Vol. 25, No. 5, pp. 3317–3336

A systematic survey and taxonomy of run-time error detection and introspection mechanisms for deep neural perception in automated vehicles.

DOI

BibTeX

@article{yatbaz2024tits,
  title   = {Introspection of {DNN}-Based Perception Functions in Automated Driving Systems: State-of-the-Art and Open Research Challenges},
  author  = {Yatbaz, Hakan Yekta and Dianati, Mehrdad and Woodman, Roger},
  journal = {IEEE Transactions on Intelligent Transportation Systems},
  volume  = {25},
  number  = {5},
  pages   = {3317--3336},
  year    = {2024},
  doi     = {10.1109/TITS.2023.3315070}
}

Warwick PhD 2024 PhD ThesisDoctoral Thesis

Runtime Error Detection for Deep Neural Network-Based Object Detection for Perception in Automated Driving Systems

H. Y. Yatbaz (Supervised by Prof. Mehrdad Dianati, Dr. Roger Woodman)

PhD Thesis · WMG, University of Warwick, UK

Doctoral dissertation establishing multi-layer activation introspection and spatial filtering for 2D and 3D perception in automated driving.

Warwick WRAP

BibTeX

@phdthesis{yatbaz2024thesis,
  title  = {Runtime Error Detection for Deep Neural Network-Based Object Detection for Perception in Automated Driving Systems},
  author = {Yatbaz, Hakan Yekta},
  school = {University of Warwick},
  year   = {2024}
}

2023

ICCVW 2023 ConferenceAutonomous Driving

Introspection of 2D Object Detection using Processed Neural Activation Patterns in Automated Driving Systems

H. Y. Yatbaz, M. Dianati, K. Koufos, R. Woodman

IEEE/CVF International Conference on Computer Vision Workshops (ICCVW), ROAD++ Workshop, Paris, France

Introduces processed neural activation pooling to monitor 2D object detectors for missed detections in road scenes.

DOI IEEE Xplore ResearchGate

BibTeX

@inproceedings{yatbaz2023iccvw,
  title     = {Introspection of 2D Object Detection using Processed Neural Activation Patterns in Automated Driving Systems},
  author    = {Yatbaz, Hakan Yekta and Dianati, Mehrdad and Koufos, Konstantinos and Woodman, Roger},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops (ICCVW)},
  year      = {2023},
  doi       = {10.1109/ICCVW60793.2023.00437}
}

IoT Elsevier 2023 JournalE-Health / IoT

A Smart E-Health Framework for Monitoring the Health of the Elderly and Disabled

A. Yazıcı, D. Zh. Zhumabekova, A. Nurakhmetova, Z. Yergaliyev, H. Y. Yatbaz, Z. Makisheva, M. I. Lewis, E. Ever

Internet of Things (Elsevier), Vol. 24, Art. 100971

An end-to-end IoT and edge-sensing architecture for continuous anomaly and health monitoring of vulnerable individuals.

DOI

BibTeX

@article{yazici2023smart,
  title   = {A Smart E-Health Framework for Monitoring the Health of the Elderly and Disabled},
  author  = {Yaz{\i}c{\i}, A. and Zhumabekova, D. Zh. and Nurakhmetova, A. and Yergaliyev, Z. and Yatbaz, H. Y. and Makisheva, Z. and Lewis, M. I. and Ever, E.},
  journal = {Internet of Things},
  volume  = {24},
  pages   = {100971},
  year    = {2023},
  doi     = {10.1016/j.iot.2023.100971}
}

2022

ACM TOMM 2022 JournalWireless Sensing

An Effective Forest Fire Detection Framework Using Heterogeneous Wireless Multimedia Sensor Networks

B. Kizilkaya, E. Ever, H. Y. Yatbaz, A. Yazıcı

ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), Vol. 18, No. 4, Art. 108

Combines visual and scalar environmental sensors in heterogeneous wireless multimedia networks for low-false-alarm wildfire detection.

DOI

BibTeX

@article{kizilkaya2022forest,
  title   = {An Effective Forest Fire Detection Framework Using Heterogeneous Wireless Multimedia Sensor Networks},
  author  = {Kizilkaya, B. and Ever, E. and Yatbaz, H. Y. and Yaz{\i}c{\i}, A.},
  journal = {ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM)},
  volume  = {18},
  number  = {4},
  pages   = {1--21},
  year    = {2022},
  doi     = {10.1145/3473037}
}

IEEE ITSC 2022 ConferenceAutonomous Driving

Detecting Hazardous Events: A Framework for Automated Vehicle Safety Systems

D. Xiao, W. G. Geiger, H. Y. Yatbaz, M. Dianati, R. Woodman

25th IEEE International Conference on Intelligent Transportation Systems (ITSC), Macau, China

A safety monitoring framework for automated vehicles that identifies hazardous operational events and triggers fallback behaviours.

DOI

BibTeX

@inproceedings{xiao2022detecting,
  title     = {Detecting Hazardous Events: A Framework for Automated Vehicle Safety Systems},
  author    = {Xiao, D. and Geiger, W. G. and Yatbaz, H. Y. and Dianati, M. and Woodman, R.},
  booktitle = {25th IEEE International Conference on Intelligent Transportation Systems (ITSC)},
  year      = {2022},
  doi       = {10.1109/ITSC55140.2022.9921988}
}

MAMS 2022 JournalStructural ML

Prediction of Punching Shear Strength of Slab-Column Connections: A Comprehensive Evaluation of Machine Learning and Deep Learning Based Approaches

S. Derogar, C. Ince, H. Y. Yatbaz, E. Ever

Mechanics of Advanced Materials and Structures (Taylor & Francis), Vol. 31, No. 17, pp. 3888–3904

Benchmarks machine learning and deep learning models for predicting structural punching shear strength with high generalisation.

DOI

BibTeX

@article{derogar2022prediction,
  title   = {Prediction of Punching Shear Strength of Slab-Column Connections: A Comprehensive Evaluation of Machine Learning and Deep Learning Based Approaches},
  author  = {Derogar, S. and Ince, C. and Yatbaz, H. Y. and Ever, E.},
  journal = {Mechanics of Advanced Materials and Structures},
  volume  = {31},
  number  = {17},
  pages   = {3888--3904},
  year    = {2022},
  doi     = {10.1080/15376494.2022.2134950}
}

ICISCT 2022 ConferenceE-Health / ML

Critical Analysis of Validation Methods for Machine Learning Models Used in E-Health Applications

H. Y. Yatbaz, A. Yazıcı, E. Ever

International Conference on Information Science and Communications Technologies (ICISCT), Tashkent, Uzbekistan

Evaluates cross-validation pitfalls and data leakage in sensor-based e-health machine learning pipelines.

DOI

BibTeX

@inproceedings{yatbaz2022critical,
  title     = {Critical Analysis of Validation Methods for Machine Learning Models Used in E-Health Applications},
  author    = {Yatbaz, H. Y. and Yaz{\i}c{\i}, A. and Ever, E.},
  booktitle = {International Conference on Information Science and Communications Technologies (ICISCT)},
  year      = {2022},
  doi       = {10.1109/ICISCT55600.2022.10146901}
}

2021

IEEE Sensors 2021 JournalE-Health / CNN

Activity Recognition and Anomaly Detection in E-Health Applications Using Color-Coded Representation and Lightweight CNN Architectures

H. Y. Yatbaz, E. Ever, A. Yazıcı

IEEE Sensors Journal, Vol. 21, No. 11, pp. 12799–12810

Encodes heterogeneous sensor streams into compact 2D colour-coded images for lightweight CNN anomaly detection.

DOI

BibTeX

@article{yatbaz2021activity,
  title   = {Activity Recognition and Anomaly Detection in E-Health Applications Using Color-Coded Representation and Lightweight {CNN} Architectures},
  author  = {Yatbaz, Hakan Yekta and Ever, Enver and Yaz{\i}c{\i}, Adnan},
  journal = {IEEE Sensors Journal},
  volume  = {21},
  number  = {11},
  pages   = {12799--12810},
  year    = {2021},
  doi     = {10.1109/JSEN.2021.3061458}
}

2020

IEEE IWBF 2020 ConferenceBiometrics / Forensics

Deep Learning Based Stress Prediction From Offline Signatures

H. Y. Yatbaz, M. Erbilek

8th International Workshop on Biometrics and Forensics (IWBF), Porto, Portugal

Investigates deep visual representations for predicting human psychological stress from offline handwritten signatures.

DOI

BibTeX

@inproceedings{yatbaz2020stress,
  title     = {Deep Learning Based Stress Prediction From Offline Signatures},
  author    = {Yatbaz, Hakan Yekta and Erbilek, M.},
  booktitle = {8th International Workshop on Biometrics and Forensics (IWBF)},
  year      = {2020},
  doi       = {10.1109/IWBF49977.2020.9107942}
}

IEEE SIU 2020 ConferenceSignal Processing

Exploring Activity Transitions in the STA-based Activity Recognition

S. Eraslan, H. Y. Yatbaz, E. Ever, Y. Yeşilada

28th Signal Processing and Communications Applications Conference (SIU), Gaziantep, Turkey

Models sequential transitions between daily living activities using Scanpath Trend Analysis over ambient sensor events.

DOI

BibTeX

@inproceedings{eraslan2020exploring,
  title     = {Exploring Activity Transitions in the {STA}-based Activity Recognition},
  author    = {Eraslan, S. and Yatbaz, H. Y. and Ever, E. and Ye{\c{s}}ilada, Y.},
  booktitle = {28th Signal Processing and Communications Applications Conference (SIU)},
  year      = {2020},
  doi       = {10.1109/SIU49456.2020.9302074}
}

IET Chapter 2020 Book ChapterSensory Analytics

Statistical Analysis for Sensory E-Health Applications: Opportunities and Challenges

H. Y. Yatbaz, A. C. Tasıran

Book chapter in Sensory Data Analysis for Health Monitoring, Institution of Engineering and Technology (IET), pp. 165–186

Reviews statistical modelling methods and reliability challenges for noisy continuous health-monitoring sensor data.

DOI

BibTeX

@incollection{yatbaz2020statistical,
  title     = {Statistical Analysis for Sensory E-Health Applications: Opportunities and Challenges},
  author    = {Yatbaz, Hakan Yekta and Tas{\i}ran, A. C.},
  booktitle = {Sensory Data Analysis for Health Monitoring},
  publisher = {Institution of Engineering and Technology (IET)},
  pages     = {165--186},
  year      = {2020},
  doi       = {10.1049/pbce128e_ch7}
}

2019

IEEE Sensors 2019 JournalAmbient Sensing / IoT

Activity Recognition Using Binary Sensors for Elderly People Living Alone: Scanpath Trend Analysis Approach

H. Y. Yatbaz, S. Eraslan, Y. Yeşilada, E. Ever

IEEE Sensors Journal, Vol. 19, No. 18, pp. 8249–8260

Adapts eye-tracking Scanpath Trend Analysis (STA) to recognise daily activities and anomalies from binary ambient home sensors.

DOI

BibTeX

@article{yatbaz2019activity,
  title   = {Activity Recognition Using Binary Sensors for Elderly People Living Alone: Scanpath Trend Analysis Approach},
  author  = {Yatbaz, Hakan Yekta and Eraslan, Sukru and Ye{\c{s}}ilada, Yeliz and Ever, Enver},
  journal = {IEEE Sensors Journal},
  volume  = {19},
  number  = {18},
  pages   = {8249--8260},
  year    = {2019},
  doi     = {10.1109/JSEN.2019.2915026}
}

2018

IEEE LCNW 2018 ConferenceMobile Networks

Hybrid Approach for Disaster Recovery using P2P Communications in Android

H. Y. Yatbaz, B. Cinar, A. Gökdemir, E. Ever, F. M. Al-Turjman, H. X. Nguyen, A. Yazıcı

43rd IEEE Conference on Local Computer Networks Workshops (LCN Workshops), Chicago, USA

A resilient peer-to-peer communication protocol on mobile devices for emergency coordination when infrastructure fails.

DOI

BibTeX

@inproceedings{yatbaz2018hybrid,
  title     = {Hybrid Approach for Disaster Recovery using {P2P} Communications in {Android}},
  author    = {Yatbaz, H. Y. and Cinar, B. and G{\"o}kdemir, A. and Ever, E. and Al-Turjman, F. M. and Nguyen, H. X. and Yaz{\i}c{\i}, A.},
  booktitle = {43rd IEEE Conference on Local Computer Networks Workshops (LCN Workshops)},
  year      = {2018},
  doi       = {10.1109/LCNW.2018.8628555}
}

06 - Research vision

Next: perception that knows its limits in shared spaces.

Where the thread leads: from single-vehicle neural activation monitors to self-assessing, cooperative multi-agent autonomy.

Dr. Hakan Yekta Yatbaz standing in front of the Queen's University Belfast autonomous research vehicle and the Lanyon Building — With the instrumented autonomous research vehicle in front of the Lanyon Building at Queen's University Belfast. School of EEECS.

Real-World Autonomy · Queen's University Belfast

From Neural Activations to Road-Ready Trust

Bridging theoretical AI safety and physical autonomous platforms

At Queen's University Belfast (School of EEECS), our research connects foundational advances in deep neural network introspection with real-world multi-modal sensing platforms. By monitoring internal activation dynamics across LiDAR, camera, and cooperative V2X streams, we aim to build autonomous systems that continuously quantify their own perceptual reliability in complex, everyday environments.

Research strands: 04Core Agendas
Doctoral supervision: DfE& CTA Funded

Open problems & collaboration PhD Studentships

Strand 01 · Sensing

Robust & Self-Calibrating 3D Sensing in Autonomy

Multi-modal perception pipelines that stay reliable under adverse weather, sensor degradation, noise, and domain shift in real-world deployments.

PhD project scope: Robust & Self-Calibrating 3D Sensing in Autonomy

Strand 02 · Introspection

Anticipatory Introspection & Self-Correcting Autonomy

Introspection frameworks and neural activation monitors that detect perception failures and out-of-distribution inputs before they become safety incidents.

PhD project scope: Anticipatory Introspection & Self-Correcting Autonomy

Strand 03 · Representations

Causally Faithful Embodied VLMs & Latent Representations

Compact activation encodings and interpretable latent representations that make neural perception transparent and accountable.

PhD project scope: Causally Faithful Embodied VLMs & Latent Representations

Strand 04 · Autonomy

Cooperative Autonomy & Anticipatory Teleoperation

Resilient collaborative perception, V2X coordination, and distributed decision-making for connected and automated vehicles and multi-agent systems.

PhD project scope: Cooperative Autonomy & Anticipatory Teleoperation

Explore funded PhD studentships & project proposals

07 - PhD & Collaborate

Open problems. Let's work on them together.

I have a ring-fenced, fully funded DfE PhD studentship available (open to UK/NI nationals, Republic of Ireland CTA citizens, and UK Settled Status / ILR holders, plus competitive international nominations) alongside academic and industry collaborations.

I
Robust & Self-Calibrating 3D Sensing in Autonomy
- Multi-modal 3D fusion
- Self-healing perception
- Domain shift & weather
II
Anticipatory Introspection & Self-Correcting Autonomy
- Failure anticipation
- Latent-steered E2E driving
- AI safety & control
III
Causally Faithful Embodied VLMs & Latent Representations
- Embodied VLMs / VLAs
- Shortcut & confounder auditing
- Compact latent probing
IV
Cooperative Autonomy & Anticipatory Teleoperation
- Cooperative V2X autonomy
- Networked teleoperation
- Human-AI shared control

Ways to work together

Prospective PhD students (Ring-Fenced DfE Slot)

100% Home fees + ~£19.2k–£20.7k/yr tax-free stipend for UK/NI, Irish (CTA), EU Settled Status, and ILR candidates (CS, EE, Physics, Maths, Eng, or industry software engineers). No formal proposal needed for an initial coffee chat.

Academic & industry partners

Joint Horizon Europe / UKRI proposals, visiting researcher exchanges, and safety-critical perception validation for automated mobility.

MSc & BEng supervision

Queen's University Belfast students interested in dissertation projects on AI safety, LiDAR perception, or XAI are welcome to reach out.

Topics, funding eligibility & how to apply

h.yatbaz@qub.ac.uk

Ashby Building, Room 08.010 · School of EEECS, Queen's University Belfast, Belfast, UK

08 - Service & background

Talks, service and background

Editorial & Peer Review

Associate Editor, Connected Mobility and Automation (Frontiers in Future Transportation).

IEEE Transactions on Intelligent Transportation Systems
IEEE Transactions on Intelligent Vehicles
IEEE Transactions on Fuzzy Systems
IEEE Sensors Journal
Scientific Reports (Nature Portfolio)

Affiliations: Associate Fellow of Advance HE (PR307078) · Member, IEEE Intelligent Transportation Systems Society.

Invited Talks & Education

Can We Trust AI in Autonomous Systems? From Perception to Run-Time Error Detection — IEEE QUB Student Branch (09 Oct 2026). Key challenges in trustworthy AI, machine-learning safety, reliable perception, computer vision, and run-time error detection for autonomous systems that make decisions in real time.
Ph.D. in Engineering, University of Warwick (2020–2024). Runtime error detection for DNN-based perception in automated driving, advised by Dr. Roger Woodman and Prof. Mehrdad Dianati.
B.Sc. and M.Sc. in Computer Engineering, Middle East Technical University (METU NCC) (2013–2020).

For hosts

Dr. Hakan Yekta Yatbaz is a Lecturer (Assistant Professor) in Autonomous Systems in the School of Electronics, Electrical Engineering and Computer Science (EEECS) at Queen's University Belfast. His research focuses on safe, reliable, and trustworthy AI for autonomous systems, spanning robust multi-modal sensing, run-time monitoring and introspection, explainable representations, and cooperative autonomy. Before joining Queen's, he was a Postdoctoral Research Fellow at WMG, University of Warwick, where he worked on the EU-funded EVENTS and Hi-Drive automated driving projects and completed his Ph.D. in Engineering. He holds B.Sc. and M.Sc. degrees in Computer Engineering from Middle East Technical University (METU NCC) and serves as an Associate Editor for Frontiers in Future Transportation.

Download headshot