[AllUsers-ISR] Palestra do Doutor João Filipe Henriques - Anfiteatro ISR, 4 de Abril, 16h00

[Lara Costa]

KIND REMINDER

 

De: allusers-bounces at lists.isr.uc.pt [mailto:allusers-bounces at lists.isr.uc.pt] Em nome de Jorge Batista
Enviada: 2 de abril de 2024 16:43
Para: docentes at deec.uc.pt; allusers at isr.uc.pt; alunos at deec.uc.pt
Assunto: [AllUsers-ISR] Palestra do Doutor João Filipe Henriques - Anfiteatro ISR, 4 de Abril, 16h00

 

 

Boa tarde colegas,

Boa tarde à comunidade estudantil do DEEC,

 

Gostava de vos convidar para uma palestra do Prof. João Filipe Henriques, antigo aluno de mestrado e de doutoramento do DEEC, que está de visita a Coimbra. 

O doutor João Henriques é Research Fellow da Royal Academy of Engineering e investigador no Visual Geometry Group na Universidade de Oxford.

 

Tópico da Palestra “Learning Location-Consistent Visual Features".

 

Dia: 04-Abril (quinta-feira)

Hora: Início às 16h00

Duração : aproximadamente 60min

Local: Anfiteatro do ISR

Mais detalhes: ver abaixo.

 

Com os melhores cumprimentos

Jorge Batista

 

Jorge Manuel M.C. Pereira Batista
Associate Professor w/ Habilitation
ISR Senior Researcher
DEEC/FCTUC
University of Coimbra
Coimbra, PORTUGAL

 

 

_________________________________________________________________________________________________

 

 

 

Dr. João Henriques is a Research Fellow of the Royal Academy of Engineering, working at the Visual Geometry Group (VGG) at the University of Oxford. His research focuses on computer vision and deep learning, with the goal of making machines more perceptive, intelligent and capable of helping people. He created the KCF and SiameseFC visual object trackers, which won the highly competitive VOT Challenge twice, and are widely deployed in consumer hardware, from Facebook apps to commercial drones. His research spans many topics: robot mapping and navigation, including reinforcement learning and 3D geometry; multi-agent cooperation and "friendly" AI; as well as various forms of learning, from self-supervised, causal, and meta-learning, including optimisation theory.

 

 

In this talk I will discuss recent work on learning location-consistent visual features, and time-permitting will also briefly discuss recent work on robotics.

 

"LoCo: Memory-Efficient Learning of Location-Consistent Features "

 

Image feature extractors are rendered substantially more useful if different views of the same 3D location yield similar features. A feature extractor that achieves this goal even under significant viewpoint changes must recognise not just the semantic categories present in a scene, but also understand how different objects relate to each other in three dimensions.

 

We present a method for memory-efficient learning of location-consistent features that reformulates and approximates the smooth average precision objective. This novel loss function enables improvements in memory efficiency by a factor of 2000, mitigating a key bottleneck of previous methods and allowing much larger models to be trained with the same computational resources.

 

"Rapid Motor Adaptation for Robotic Manipulator Arms "

 

Developing generalizable manipulation skills is a core challenge in embodied AI. This includes generalization across diverse task configurations, encompassing variations in object shape, density, friction coefficient, and external disturbances such as forces applied to the robot. 

 

Rapid Motor Adaptation (RMA) offers a promising solution to this challenge.

It posits that essential hidden variables influencing an agent's task performance, such as object mass and shape, can be effectively inferred from the agent's action and proprioceptive history. 

 

Drawing inspiration from RMA in locomotion and in-hand rotation, we use depth perception to develop agents tailored for rapid motor adaptation in a variety of manipulation tasks.

 

We evaluated our agents on four challenging tasks from the Maniskill2 benchmark, namely pick-and-place operations with hundreds of objects from the YCB and EGAD datasets, peg insertion with precise position and orientation, and operating a variety of faucets and handles, with customized environment variations.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isr.uc.pt/archive/allusers/attachments/20240404/7845440e/attachment.html>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Anexo sem nome 00640.txt
URL: <https://lists.isr.uc.pt/archive/allusers/attachments/20240404/7845440e/attachment.txt>