Job description
Are you passionate about AI, multi-modal sensing, and resilient infrastructure? Would you like to integrate advanced AI techniques with heterogeneous UAV-based sensor data for the inspection and maintenance of critical transportation infrastructure, addressing real-world challenges faced by industry, governments, and society within the international STRUCTURE project?
Information
The PhD candidate will work within the international research project STRUCTURE in cooperation with industrial partners from the Netherlands, the United Kingdom, Belgium, Turkey, and Portugal. The project aims to automate and enhance inspection of transportation infrastructure through multi-modal sensing, autonomous UAV platforms, and advanced AI-based analysis. A central focus is the detection of defects in bridges and viaducts using X-ray, LiDAR, visual and acoustic data captured from UAVs.
The research will address the design of AI models capable of combining heterogeneous sensor modalities, including RGB, thermal, LiDAR, acoustic arrays, GPR, and X-ray backscatter, to create a unified and reliable representation of structural integrity. The work expands on TU/e’s contributions by developing algorithmic components for detection and classification of defects and anomalies across both surface and subsurface layers. This includes constructing robust feature extraction pipelines, attention-based fusion architectures, and deep learning models that accurately characterize cracks, voids, delamination, corrosion, and internal structural discontinuities. The PhD candidate will investigate Vision Language Models (VLMs), Multi-modal AI solutions, and 3D scene reasoning approaches, to achieve spatial understanding and cross-modal representation learning from heterogeneous sensor data, with the research not limited to these methods. This research will support semantic interpretation, defect localization, temporal reasoning, and predictive maintenance in complex inspection environments.
A second contribution involves predictive maintenance algorithms that integrate static data sources (such as geological maps, material properties, and usage profiles) with dynamic sensor measurements (including pressure, vibration, visual, acoustic, and X-ray signals). The PhD candidate will investigate temporal modelling, multimodal analysis, and risk progression modelling to forecast deterioration patterns and estimate the remaining useful lifetime of infrastructure components. The research also encompasses model compression and optimization for edge deployment on UAV-mounted processors to support real-time inference. The candidate will collaborate with industrial partners for real-world data acquisition and large-scale validation on operational bridges and viaducts.
Research group and company
The PhD student will be working in the AIMS laboratory of the Signal Processing Systems (SPS) group within the Department of Electrical Engineering at TU/e. The AIMS lab researches and develops AI models for systems equipped with sensors of multiple different modalities. We foster expertise in AI analysis of RGB, thermal, depth, LiDAR, acoustic, sonar and radar sensor data, with established research lines in 3D reconstruction, and Edge AI for resource-constrained deployments.