How will the physical AI evolve being exposed to data from various sensors and radio signals? This is the main underlying theme to be explored within this postdoctoral position.
The appointed researcher will investigate how AI embedded in physical entities, such as robots, vehicles, or sensors, forms internal representations of space, time, and motion when interacting in complex non-stationary environments. The objective is to study and develop models for physical AI systems that learn and adapt through continuous exposure to multimodal sensory and radio data, and acts upon real-world environment through distributed coordination and control. Emphasis will be placed on how real-world communication constraints influence learning and perception, particularly in wireless and networked settings.
Specifically, the research will explore how 6G wireless networks, ubiquitous WiFi, and satellite channels can support robust perception and inference for distributed AI systems, and how delays, interference, and signal imperfections affect cognitive performance. The candidate will design models and algorithms for learning and decision-making under uncertainty, optimized for real-time operation on heterogeneous physical devices. Finally, the position will address how communication and computation co-design enables scalable interaction among networked AIs, paving the way toward embodied intelligence in large-scale physical systems.
Prospective applicants for this postdoctoral should have the following qualifications:
Appointment as postdoc requires academic qualifications at PhD level.