About Us






The Lab

The Visual Artificial Intelligence Laboratory was founded in 2012 by Professor Cuzzolin, under the name of 'Machine Learning' (and later 'Artificial Intelligence and Vision') research group, and has since been conducting work at the boundaries of human action recognition in computer vision. Prof Cuzzolin is a leading scientist in the mathematics of uncertainty, in particular random set and belief function theory. The team's interests have since expanded towards machine learning and Artificial General Intelligence, robotics, in particular surgical robotics and autonomous driving, as well as AI for Healthcare.

The team is projected to comprise around 30 members in 2021, including 1 administrator, 4 faculty members, 2 KTP associates, 11 postdocs and research fellows, 6 PhD students, 3 MSc students, 2 interns, various visitors and 4/5 external collaborators.


Research Focus

The group has built in just a few years a leadership position in the field of deep learning for action detection, which has led to be best detection accuracies to date and the only system so far able to localise multiple actions on the image plane in (better than) real time. The team's effort is now shifting towards topics at the frontier of visual AI, such as future action prediction, deep video captioning and the development of a theory of mind for machines. We also working towards applying our expertise in the area to autonomous driving.

Prof Cuzzolin's reputation in uncertainty theory and belief functions comes from the formulation of a geometric approach to uncertainty in which probabilities, possibilities, belief measures and random sets are represented and analysed by geometric means. Current exciting work includes the generalisation of the law of total probability, the notions of upper and lower likelihoods, and that of generalised random variables.
A 10-year summary of this work on the geometry of uncertainty is being released as a Springer monograph in December 2020.
This has recently developed into an effort to reshape the foundations of artificial intelligence to better incorporate and model second-order, 'epistemic' uncertainty: an approach that we call Epistemic Artificial Intelligence.

Within machine learning the team's work is directed at understanding the mathematics of deep learning, providing new robust foundations for statistical learning theory, and developing novel tools based on the theory of random sets, in particular generalisation of the logistic regression frameworks and of max-entropy classifiers. We are part of research consortia which aim at applying AI to human-robot interaction (the creation of emotional avatars), surgical robotics (robotic assistant surgeon arms) and healthcare (the early diagnosis of dementia and the monitoring of people in a coma).




Funding

The Lab currently runs on a budget of around £2.7M (not fully incorporating the €4.3M Horizon 2020 project SARAS), with currently seven live projects funded by Horizon 2020 (2), the Leverhulme Trust, Innovate UK, Huawei Technologies, UKIERI, and the School of Engineering, Computing and Mathematics. The group receives annually QR funding from the Faculty for around £7,000. The budget is projected to further significantly increase in 2021.

A number of both international and local partner companies are both supporting our research bids or considering joint grant applications or direct funding.




Facilities

For our computation-intensive research we rely on a network of several GPU-powered workstations: (1) Mercury - 4 GPUs; (2) Mars - 4 GPUs, 5 HDD disks (18.5 TB), 1 SSD Disk (1TB); (3) Sun - 2 GPUs, 4 HDD disks (12.5 TB); (4) Jupiter - 1 GPU, 2 HDD disks (4.2 TB); (5) Earth - 2 HDD disks (4TB); (6) Venus - 1 HDD (0.5TB). We also possess two CPU servers with 48 CPU cores and 128GB RAM each, and have access to the Faculty's High Performance Computer.

A new £33K 8-GPU workstation equipped with 192Gb RTX cards funded by the Faculty of Technology, Design and Environment has recently been purchased.

Two more high-end 8-GPU servers, funded by SARAS and Epistemic AI, are in the process of being purchased.




Robotics & Fab Lab

The adjacent Cognitive Robotics Lab is well-equipped, with 1 Baxter robot with AR10 hand, vacuum and parallel grippers, 1 Robothespian (Artie), 4 Naos, 2 in-house designed and built humanoid robots (Blu), 5 AX-18A Smart Robotic Arms, 6 humanoid Minis. The Fabrication lab is endowed with 5 Aunar Upgraded Desktop and 2 Raise3D N2 Plus 3D Printers. The Performance Augmentation Lab is in possession of 2 Microsoft Hololens and 2 Epson BT200 & BT2000.




Partners

The Visual AI Lab is part of the Institute for Ethical AI, and collaborates very closely with a number of research groups and companies: Oxford University's Torr Vision Group (formerly known as Brookes Vision Group), the Centre for Movement, Occupational and Rehabilitation Sciences (MOReS) led by Prof Dawes, the Autonomous Driving group led by Dr Bradley, Thomas Lukasiewicz, Professor at Oxford University's Department of Computer Science, the Neuroscience group at Cambridge University, the Performance Augmentation Lab led by Dr Wild, and the Centre for Biomedical Cybernetics of the University of Malta, led by Prof Camilleri.

Industrial partners include Huawei Technologies, Sportlight/Createc, Peoplespace and Disney Research, among others.