Triple helix for control: model based learning and embodied intelligence
Often robotics and artificial intelligence are seen as the same. But many robots do not use artificial intelligence algorithms for the control of their behaviour. Artificial intelligence systems provide robotics application new impressive perspectives, but also come with complex challenges.
Praktische info:
Inschrijven?
- Prijs: gratis
Webinars vinden plaats één keer per maand. Registratie is niet nodig, je kan zo naar het seminarie komen.
Krijg je graag een maandelijkse herinnering aan de seminaries in deze reeks over tijdsreeksen? Schrijf je in!
Georganiseerd door:
Often robotics and artificial intelligence are seen as the same. But many robots do not use artificial intelligence algorithms for the control of their behaviour. Their control is dependent on known models, and, based on these models, calculations and input data, a complex algorithm defines the output and hence the behaviour of the robot.
However, as such the system cannot learn. Artificial intelligence systems provide robotics application new impressive perspectives, but also come with complex challenges. For aspects like vision and speech recognition, deep learning has drastically outperformed traditional methods. These approaches typically require massive amounts of labelled data and computational power. Although data is typically abundant in robotics, labelling is sparse and expensive. For the control of robots, reinforcement learning often requires significantly more iterations than are feasible on real systems.
Therefore, a lot of the learning work is done in virtual environments, with challenges to transfer to the real world. Moreover, robots are interacting with the real world and challenges exist how to learn in a safe way, for the humans, for the environment and for the robot itself.
Both methods, model-based control and learning methods, have disadvantages and advantages, and combining the best of both worlds is probably an interesting solution. Moreover like in animals and humans, due to millions years of evolution, the bodies are optimized for the tasks and the environment. Through a smart design with the use of new materials, actuators and sensor, part of the computational intelligence can be outsourced to the embodied intelligence of the robot body.
According to the paradox of Moravec “AI can quickly learn to perform tasks considered "difficult" by humans”, such as complex statistics and analysis. However, on the other hand, tasks that are trivial for humans, are often still exceedingly difficult for computers and robots.” Therefore instead of replacing humans by robots, there has been a lot of focus on developing collaborative robots to exploit these complementary strengths.
Lesgever/spreker
Bram Vanderborght
Prof. dr. ir. Bram Vanderborght was born in Belgium in 1980. He received the degree in the study of Mechanical Engineering at the Vrije Universiteit Brussel in 2003 with highest distinction. Since 2003 he was researcher at the VUB, supported by the Fund for Scientific Research Flanders (FWO). In May 2007 he received his PhD in Applied Sciences. The focus of his research was the use of adaptable compliance of pneumatic artificial muscles in the dynamically balanced biped Lucy. In May-June 2006 he performed research on the humanoids robot HRP-2 at the Joint Japanese/French Robotics Laboratory (JRL) in AIST, Tsukuba (Japan) in the research "Dynamically stepping over large obstacles by the humanoid robot HRP-2". He received a 3-year post-doc grant with mobility grant from the FWO. From October 2007-April 2010 he worked as post-doc researcher at the Italian Institute of Technology in Genova (Italy) on the humanoid robot iCub and compliant actuation.
Since October 2009, he is appointed as professor at the VUB where he teaches mechatronics and gives a robotics project. From October 2011-Sept 2016, he was research director at the Universitatea Babes-Bolyai, Department of Clinical Psychology and Psychotherapy with a project on robot assisted therapy with ASD children and VUB-PI of the DREAM project. He received an ERC Starting Grant on Series-Parallel Elastic Actuation for Robotics (SPEAR). He was a member of the Young Academy of the Flemish Academy of Belgium for Science and the Arts.
Currently he coordinates EU FET SHERO SHERO project and Marie Curie ITN SMART project. He is VUB-PI of EU project SOPHIA. He was till 2020 the core lab manager of Flanders Make in Flexible Assembly, now he is affiliated to the Interuniversity Microelectronics Institute (imec), Belgium, as scientific collaborator. He is also member of Brubotics, the mulitdisciplinary research enter on Human robot interaction. He also co-leads the Homo Roboticusproject on how to keep the human values central in a robotised world. He was till 2020 the Editor In Chief of IEEE Robotics and Automation Magazine, from 2022 he will be Vice President - Electronic Products and Services Board of IEEE-RAS.
His research interests include the use of soft and self healing actuators for cognitive and physical human robot interaction in applications in health and manufacturing. His works is on display in the AI Experience Center.
Machine Learning in Control and Automation Webinar Series
In this webinar series, we bring together researchers that are interested in, or conducting research on Machine Learning for control and automation purposes. A variety of techniques and their applications will be covered, ranging from traditional machine learning techniques such as System Identification, State Estimation and Model Predictive Control (MPC) to Deep Neural Networks (DNN)-based approaches and Reinforcement Learning. We offer a varied program of national and international speakers.
Blijf op de hoogte van dit onderwerp / deze reeks
Gerelateerde opleidingen
Interdisciplinary School on Machine Learning and AI for Science
Interdisciplinary school - Heilbronn - ETH Zürich
Machine Learning Specialization
opleiding - online - DeepLearning.AI, Stanford Online