¹û¶³´«Ã½

Egor Bondarev is an Associate Professor chairing the ‘AI for Multi-modal Sensing’ (AIMS) lab within the ‘Signal Processing Systems’ group of ¹û¶³´«Ã½. His AIMS lab focuses on research of AI/ML models for multi-modal perception systems in the smart city, public safety, intelligent transportation, defense and broadcast domains. The developed AI models perform analysis by fusing data of the following sensor modalities: thermal, LiDAR, RGB, depth, acoustic and 4D imaging radar.  The overarching objective of this fusion-empowered analysis is to provide real-time Situational Awareness to the abovementioned industries.

Bondarev is focusing on two key research problems: 1) how the subtle anomalies in human behaviour or industrial assets can be detected without having training datasets of anomaly samples; 2) how to fuse the 2D-data from individual multi-modal sensors capturing a scene into an accurate 3D spatio-temporal model of the scene, thereby bringing real-time situational awareness. His additional research contributions are sparsification and mapping of the AI models on edge devices and explainable AI for privacy preservation.

Egor Bondarev is involved as principal investigator in several European research projects, such as SMART Mobility, SINTRA, ELEVATION and ADVISOR. He organized valorization and exploitation of the AI Models developed in such projects at ST Microelectronics (France), ViNotion (Netherlands), INRIA (France) and ProDrive (Netherlands). His extensive collaboration network includes both academia (TU Munich, Gent University, Birmingham University, etc) and industry (Philips, NXP, ASML, Bosch, Demcon, Rijkswaterstaat, Esri Canada, TAV Turkey, etc)

As a lecturer, Bondarev is responsible for the master course ‘Computer Vision AI and 3D Data Processing’ and also teaches the bachelor course Computer Architecture I. He has written and co-authored 100+ publications on real-time computer vision, AI and image/3D processing algorithms. He is the organizer of the ePictureThis workshop series, ICMV and ACIVS conferences.