The research profile sits at the intersection of computer vision, artificial intelligence, and cognitive robotics. It focuses on semantic representations of 2D and 3D visual experiences, building on the idea that structured symbolic descriptions of observed scenes can support robust perception and intelligent action execution in autonomous systems, including robots, intelligent vehicles, and other embodied AI platforms.

A central focus is helping autonomous systems understand their surroundings and act reliably. The research spans multimodal semantic scene understanding, multitask learning, the detection and anticipation of robot failures, and perception in challenging environmental conditions. Current national and internation projects explore combining data from sensors such as 3D LiDAR, 4D RADAR, and RGB-D cameras to improve situational awareness and system performance in difficult conditions, including extreme weather.