Research
Three interconnected research themes and four dedicated programs that translate industrial embodied intelligence into deployable systems—from dexterous manipulation and physics-informed manufacturing to human–machine collaboration.
RAIDS builds on three interconnected pillars in AI for industrial digital servitization.
Mutual-cognitive collaboration, vision-language action models, mixed reality, and safe motion generation for human-centric smart manufacturing.
Digital twins, intelligent design, and service-dominant logic toward adaptive, data-driven industrial product-service solutions.
Embodied AI, contact-rich robot learning, and physics-informed methods that bridge perception, cognition, and execution on the factory floor.
Dedicated tracks translating our themes into deployable systems, platforms, and field demonstrations.
AeroDexter co-designs micro-UAVs with lightweight manipulators for contact-rich dexterous tasks in hard-to-reach spaces—unifying flight dynamics, force control, and embodied learning toward terrestrial-grade manipulation in free three-dimensional environments.
COOLAND advances intelligent manufacturing for aerospace and high-end equipment through Physics-Informed AI and real-time physical field modeling. An operator-centric Human-in-the-Loop framework, paired with Intelligent Flexible Manufacturing, enables agile reconfiguration of assembly and machining lines.
Glovity develops precise, dexterous manipulation policies for contact-rich industrial tasks, and translates them through TeleX—proprietary egocentric sensing, multimodal capture, and learning pipelines that turn expert skill into deployable autonomy.
HumVe advances human–vehicle interaction for advanced driver assistance systems through large foundation models and cognitive reasoning. A human-centric AI framework, paired with real-time multimodal perception, enables vehicles to understand driver intent and environmental context, ensuring seamless and safe collaboration on the road.
