JOB DESCRIPTION

As robots become more autonomous, the need for efficient and scalable decision-making algorithms grows. Active inference is a promising framework that unifies perception, action, and learning under a principled probabilistic model. This PhD project explores how to embed active inference into resource-constrained devices to enable responsive, adaptive manipulation in robotic systems.
The project focuses on designing robust latent spaces for predictive models, leveraging encoder-decoder and attention-based architectures for scalability, and compressing model representations through hierarchical structures—all while optimizing for real-time performance on embedded hardware.
You’ll be working at TU Delft in a collaborative, interdisciplinary environment, co-supervised by leading experts from both academia and industry. The project includes secondments at innovative partner institutions like XOsight, where you’ll apply and validate your models on real-world robotics platforms.

This position also includes two unique research visits:

• 3 months visit to XOsight, to design a scalable model to speed up large dataset processing in indoor robot inventory.
• 4 months visit to UMass, to desgin compressed yet accurate on-device active inference model.

Your home base will be the Embedded Systems group at TU Delft, where you’ll be supervised by Prof. K. Langendoen, Chair of the group, and Dr. Qing Wang, co-founder of TU Delft’s Embedded AI lab and Project Coordinator of ANT. You will be also supervised by Prof. M. Wisse, Chair of the Department of Cognitive Robotics of TU Delft, and Chair of the Robot Dynamics Group.
As a member of this dynamic and dedicated research group, you will collaborate closely with academic staff, fellow PhD candidates, and postdoctoral researchers, all committed to advancing the field of network science. Our team fosters an environment of open discussion, knowledge exchange, and regular collaboration to explore complex networks across technological, biological, and social domains. Throughout your PhD journey, you will also receive comprehensive training to support your development as a researcher.

JOB REQUIREMENTS

Required expertise:
We are looking for candidates with a background in at least one of the following domains in the context of embedded systems, robotics, and AI:

• A solid hands-on experience of embedded systems
• Embedded machine learning
• Robot operating system
• Signal processing and applied optimization

General requirements:

• Good track record in Master’s and Bachelor’s degrees
• Strong analytical or implementation skills
• Strong programming skills in one or more of the following languages: C/C++/Rust, Python
• Good command of the English language (written and spoken)

Conditions of international mobility of researchers:
Researchers are required to undertake trans-national mobility (i.e., move from one country to another) when taking up the appointment. At the time of selection by the TU Delft, the candidate must not have resided or carried out their main activity (work, studies, etc.) in the Netherlands for more than 12 months in the 3 years immediately prior to the recruitment. Short stays, such as holidays, are not taken into account.

Please refer the Job description for details