Perception Through Anticipation
The MPS lab specializes in the following AI methods: deep learning, multi-modal computer vision, and simultaneous localization and mapping. These are key enabling technologies that allow mobile sensor platforms to perceive and interpret their environments from past and current sensory data, in essence estimating a dynamic digital world-model in real-time. In coming years, we aim to make a step in the direction of spatio-temporal reasoning engines that allow mobile sensor platforms to predict possible future events and thereby achieve anticipation capabilities. Currently, the lack of anticipation capabilities, is a key bottleneck in deploying mobile autonomous systems in complex and dynamic environments, such as self-driving cars in crowded inner cities. We firmly believe that in order to advance AI and its applications, both an inter-disciplinary approach and a strong cooperation with industry are required. Hence, we are involved in many cross-disciplinary European projects and initiatives, such as the International Connected and Automated Driving Institute (https://icadi.net), and we recently started the company AI In Motion to bring our technologies to the market (https://aiim.ai).
News
Meet some of our Researchers
Recent Publications
Our most recent peer reviewed publications
-
Exploiting image translations via ensemble self-supervised learning for Unsupervised Domain Adaptation
Computer Vision and Image Understanding (2023) -
Content-aware Token Sharing for Efficient Semantic Segmentation with Vision Transformers
(2023) -
Model-free inverse reinforcement learning with multi-intention, unlabeled, and overlapping demonstrations
Machine Learning (2023) -
Intra-Batch Supervision for Panoptic Segmentation on High-Resolution Images
(2023) -
Empirical Generalization Study: Unsupervised Domain Adaptation vs. Domain Generalization Methods for Semantic Segmentation in the Wild
(2023)
Contact
-
Visiting address
FluxDe Groene Loper 195612 AP EindhovenNetherlands -
Visiting address
FluxDe Groene Loper 195612 AP EindhovenNetherlands -
Postal address
Department of Electrical EngineeringP.O. Box 5135600 MB EindhovenNetherlands -
Postal address
Department of Electrical EngineeringP.O. Box 5135600 MB EindhovenNetherlands -
Teamlead
Teamleadg.dubbelman@ tue.nl -
SecretaryVCA.secretariat@ tue.nl