PhD Position in Physical AI: Adaptive Foundation Models for Robotics

DTU — Technical University of Denmark; Favrskov Kommune; In loco; DKK 357.000 - 402.000; Tempo pieno

Descrizione del lavoro

Aarhus University invites applications for a three-year PhD fellowship in Physical AI: Adaptive Foundation Models for Robotics within the Electrical and Computer Engineering program.

The successful candidate joins the A3 Lab under senior supervision, starting 1 January 2027 or later, and will pursue original research on multimodal perception, planning and edge intelligence.

Competenze

• A master's degree in computer science, electrical or computer engineering, robotics, machine learning or a related field.
• Strong foundations in machine learning, linear algebra, probability and optimization.
• Strong Python and PyTorch skills; hands-on DL implementation experience.

Mansioni

• Conduct original Research in Physical AI for robotics.
• Develop and evaluate learning algorithms on public datasets and simulations.
• Publish results in top ML, CV and robotics venues and contribute to code repositories.

Conoscenze

Python

PyTorch

Machine learning

Deep learning

Research potential

Formazione

Master's degree in CS/EE/Robotics/ML

Descrizione del lavoro

Applicants are invited for a PhD fellowship/scholarship at Graduate School of Technical Sciences, Aarhus University, Denmark, within the Electrical and Computer Engineering programme. The position is available from 01 January 2027 or later.

TitlePhD Position in Physical AI: Adaptive Foundation Models for Robotics

Research area and project description: Applications are invited for a three-year PhD position in Physical AI at Aarhus University’s Department of Electrical and Computer Engineering. The candidate will join the Adaptive & Agentic AI (A3) Lab, supervised by Associate Professor Behzad Bozorgtabar and co-supervised by Professor Qi Zhang.

Research vision

How can intelligent robots understand instructions, anticipate the consequences of actions, and adapt reliably when the physical world changes? The project connects multimodal perception, reasoning and action with predictive learning and edge intelligence.

Research directions and objectives

Vision-language-action models (VLAs)

Investigate VLA models and multimodal representations that connect visual observations and language instructions to robot behaviour, including learning from demonstrations and generalisation to unfamiliar tasks, objects or environments.

World models and planning

Develop models that predict how the physical world responds to actions, supporting planning and learning from interaction. Possible directions include learning from video, demonstrations and simulation, and transferring knowledge across robot configurations.

Adaptation and edge intelligence

Develop efficient methods for maintaining reliable behaviour under changing environments, sensing conditions and resource constraints. Topics may include test-time and continual adaptation, uncertainty-aware decision-making, and efficient inference and model updates under latency, memory and energy limits.

The precise research focus will be developed with the successful candidate within these connected directions. Research will emphasise new learning algorithms and rigorous evaluation using public datasets, simulation and, where available, robotic and edge-computing platforms. Evaluation will consider task success, generalisation, reliability and computational efficiency. The goal is original research for leading machine-learning, computer-vision and robotics venues. The successful candidate will be expected to conduct original, high-quality research targeting publications at top-tier machine learning, computer vision and robotics conferences, including NeurIPS, ICML, ICLR, CVPR, ICCV, CoRL, RSS and ICRA.

Project description

For technical reasons, you must upload a project description. Please simply copy the project description above and upload it as a PDF in the application.

Qualifications and specific competences

Essential qualifications

Essential qualificationsA master's degree (120 ECTS or equivalent), completed by enrolment, in computer science, electrical or computer engineering, robotics, machine learning or a related field; strong academic results and foundations in machine learning, linear algebra, probability and optimisation; and strong Python and PyTorch (or comparable framework) skills.

Only applicants who demonstrate substantial hands-on experience implementing, training and evaluating deep-learning models will be considered. General interest in AI or experience limited to running tutorials or using pretrained-model APIs is not sufficient.

Applicants must provide evidence of research potential through a substantial thesis, research project, code contribution or publication, clearly identifying their own technical contribution.

Desirable experience

Research experience in multimodal foundation models or robot learning is particularly desirable, especially involving vision-language-action models or world models. Experience in model adaptation or efficient edge deployment is an additional advantage. We particularly value substantial technical contributions, demonstrated through a thesis, research project, code contribution or publication. Prior publications and hands-on robotics experience are advantageous but not required.

Application and selection

Include a one-page statement of interest identifying relevant research directions and your strongest technical project; a CV with a project portfolio and publications, where applicable; and Bachelor's and Master's transcripts and diplomas, as available. Provide links or an accessible description of relevant code, methods and results, and explain your personal contribution.

Shortlisted applicants should be prepared to discuss their own implementation, experimental design, results and limitations in a technical interview. Clear scientific communication in English and a commitment to reproducible research are essential.

Use the project description from this announcement; a separate research proposal is not required.

Place of employment and place of work

The place of employment is Aarhus University, and the place of work is Adaptive & Agentic AI (A3) Lab, Department of Electrical and Computer Engineering (ECE), Faculty of Technical Sciences, Aarhus University, Finlandsgade 22, 8200 Aarhus N, Denmark.

Contacts

• Behzad Bozorgtabar, (main supervisor)
• Qi Zhang, (co-supervisor)

Application deadline 01 November 2026 at 23:59 CET.

Preferred starting date is 01 January 2027.

Please note

• Only documents received prior to the application deadline will be evaluated. Thus, documents sent after deadline will not be considered.
• The programme committee may request further information or invite the applicant to attend an interview.
• Shortlisting will be used, which means that the evaluation committee only will evaluate the most relevant applications.

Aarhus University's ambition is to be an attractive and inspiring workplace for all and to foster a culture in which everyone has opportunities to thrive, achieve and develop. We view equality and diversity as assets, and we welcome all applicants. All interested candidates are encouraged to apply, regardless of their personal background.

Salary and terms of employment are in accordance with applicable collective agreement.