PhD Researcher on Resource-Efficient Robotic World Models

Ghent University and imec are offering a fully funded PhD position within the IDLab research group and its DECIDE team. The successful candidate will research and develop efficient robotic foundation models, such as Vision-Language-Action (VLA) models and world models, optimized for deployment on resource-constrained embedded hardware. The role involves combining techniques like quantization, pruning, and distillation to achieve real-time robot control and navigation. In addition to cutting-edge research, the candidate will contribute to academic publications and mentor students.

• Investigate and extend state-of-the-art robot control models, focusing on Vision-Language-Action models and world models for navigation and multi-step task execution.
• Implement model optimizations such as quantization, pruning, and distillation to enable deployment on embedded hardware.
• Apply developed algorithms and models to real-world robotic use cases.
• Write high-quality scientific publications targeting top-tier international journals and conferences.
• Mentor and supervise bachelor and master thesis students working on related topics.