International Workshop of Intelligent Autonomous Learning Systems
The International Workshop on Intelligent Autonomous Learning Systems is an annual, invitation-only event centered on advancements in machine learning for robotics and autonomous systems. Its primary objectives include fostering collaboration, encouraging scientific discussions, and enhancing participants' comprehension. Traditionally hosted in the picturesque Kleinwalsertal, Austria, amidst the stunning Alps.
International Workshop of Intelligent Autonomous Learning Systems 2023
Note, Please ensure to read the conference schedule and specific activity requirements before packing for the event.
Quick Facts
| Organizers: | Kai Ploeger, Kay Hansel, Ali Younes, Snehal Jauhri |
| Location: | Darmstädter Haus and Sporthotel Walliser, Kleinwalsertal, Austria |
| Dates: | 13.08.- 20.08.2023 |
Schedule
Note, the schedule is subject to change, to optimize the weather during outdoor activites.
Daily Schedule
| 13/8 | Sunday: | Travel. |
|---|---|---|
| 14/8 | Monday: | Opening session, N&R speed dating, Spotlight #1, Poster session |
| 15/8 | Tuesday: | Spotlight #2, Contributed Talks (partly parallel) |
| 16/8 | Wednesday: | Excursion (Hike or Klettersteig) |
| 17/8 | Thursday: | Spotlight #3-#4, Focus Group |
| 18/8 | Friday: | Focus Group, Group-specific session |
| 19/8 | Saturday: | Career Seminar, Grant planning & proposal writing, Closing session |
| 20/8 | Sunday: | Travel back |
Detailed Schedule
Note, there will be 30 minute coffee breaks each morning and afternoon, between sessions.
| Time | Monday | Tuesday | Wednesday | Thursday | Friday | Saturday |
|---|---|---|---|---|---|---|
| 09 - 10 | Opening Session | Spotlight #2 | Excursion | Spotlight #3 | Focus Group | Career Seminar: Layer 1 and 2 |
| 10 - 11 | N&R Speed Dating | Spotlight #2 | Excursion | Spotlight #3-#4 | Focus Group | Career Seminar: Layer 3 |
| 11 - 12 | N&R Speed Dating | Spotlight #2 | Excursion | Spotlight #4 | Focus Group | Career Seminar - Fill this form during Layer 3 when Jan asks you to! |
| 12 - 14 | Lunch | Lunch | Excursion | Lunch | Lunch | Lunch |
| 14 - 15 | Spotlight #1 | Contributed Talks | Excursion | Focus Group | Group Specific Session | Introduction to Grant Planning & Proposal Writing by Martins Zaumanis. Great book: Buy here! Useful Materials: 1 2 3 |
| 15 - 16 | Spotlight #1 | Contributed Talks | Excursion | Focus Group | Group Specific Session | Practical Grant Planning & Proposal Writing (Jan) |
| 16 - 17 | Spotlight #1 | Contributed Talks | Excursion | Focus Group | Group Specific Session | Closing Session |
| 17 - 18 | Poster Session | Contributed Talks | Excursion | Focus Group | Group Specific Session |
Invited Speakers
Nathan Lepora is a professor of Robotics and AI leading the Dexterous Robotics group at Bristol Robotics Laboratory. With a diverse academic background spanning mathematics, theoretical physics, and computational neuroscience, Nathan's research focuses on natural, artificial, and embodied intelligence in robotics and neuroscience, particularly in perception, learning, interaction, understanding, and manipulation of surroundings. Renowned for his work on robot touch and biomimetic forebrain, his research has garnered media coverage and prestigious awards, including the Leverhulme Trust Research Leadership Award. With valuable expertise in the interface of computational neuroscience, AI, and robotics, Nathan collaborates with industry leaders such as Ultraleap, Ocado Technology, Google DeepMind, and Shadow Robotics.
Jia-Jie Zhu is a research group leader at the Weierstrass Institute for Applied Analysis and Stochastics (WIAS) in Berlin. With a strong academic background, J.J. previously served as a postdoctoral researcher in machine learning at the prestigious Max Planck Institute for Intelligent Systems in Tübingen, Germany, and completed his Ph.D. in numerical optimization at the University of Florida. His expertise spans state-of-the-art machine learning, optimization, and large-scale computation, with a specific focus on algorithm development and theoretical exploration. Notably, he and his team have made significant contributions to the field, particularly in robustness theory for machine learning, utilizing computational tools such as optimization algorithms over probability distributions. Their research also delves into the fascinating intersection of large-scale computational algorithms in machine learning/optimization and dynamical systems theory, including (PDE) gradient flows and optimal transport.
Martins Zaumanis has written a book series called Peer Recognized on practical advice for writing impactful research papers, delivering powerful academic presentations, attracting research funding and making your name known on the internet. He will present an overview from his book Write a Winning Research Proposal. In his day job, he is a road pavement scientist at Empa. In his research, he strives to bring sustainability to the core of road construction! To reach this goal, he spends his days studying asphalt recycling, the use of waste materials in road construction and application warm mix asphalt technologies. In 2016, he received the International Road Federation award for the best research project of the year.
Sessions
Spotlight Talks
The spotlight talks are tutorial sessions led by field experts, providing a clear and informative introduction to a specific research field. These talks aim to acquaint the audience with essential concepts, recent developments, and significant findings in that area of study.
| Time | Talk | Title | Presenter |
|---|---|---|---|
| Monday 14:00 - 17:00 | #1 | Touch sensing and Processing in Robotics | Nathan Lepora (University of Bristol), Roberto Calandra (TU Dresden) |
| Tuesday 09:00 - 12:00 | #2 | Optimization and dynamics: from Euclidean gradient descent to Wasserstein gradient flow | J. J. Zhu (WIASS, Berlin) |
| Thursday 09:00 - 10:30 | #3 | Hierarchical Reinforcement Learning | Herke van Hoof (University of Amsterdam) |
| Thursday 10:45 - 12:15 | #4 | Generative Models for Optimization and Planning | Julen Urain (TU Darmstadt) |
Contributed Talks
During this session, researchers from different labs present their research, providing insights, methodologies, and findings on specific topics or projects while also receiving feedback from the audience.
Note, there will be two talks in parallel, denoted #1 and #2.
| Time | Talk | Title | Presenter |
|---|---|---|---|
| 14:00 - 14:30 | A1 | Interactive Learning and Generalization of Dynamical Motor Primitives | Ravi Prakash (CoR, TU Delft) |
| A2 | Leveraging Multi-Agent Reinforcement Learning and Graph Neural Networks for Distributed Traffic-Aware Routing Optimization | Andreas Boltres (ALR, KIT) | |
| 14:35 - 15:05 | B1 | Diffusion for Goal Conditioned Imitation Learning | Moritz Reuss (IRL, KIT) |
| B2 | Neural Networks as Sparse Gaussian Processes for Sequential Learning | Aidan Scannell (RL Lab, Aalto) | |
| 15:10 - 15:40 | C1 | Coherent Soft Imitation Learning | Joe Watson (IAS, TU Da) |
| C2 | Simplified Temporal Consistency Reinforcement Learning | Yi Zhao (RL Lab, Aalto) | |
| 15:45 - 16:15 | D1 | Search-based Task and Motion Planning | Zlatan Ajanovic (CoR, TU Delft) |
| D2 | Reinforcement Learning in Quadruped Robots: Tackling Varied Scenarios | Qingfeng Yao (RL Lab, Aalto) | |
| 16:20 - 16:50 | E1 | A Time-Continuous Option Framework | Samuele Tosatto (IIS, UIBK) |
| E2 | Multi Time Scale World Models | Vaisakh Shaj Kumar (ALR, KIT) | |
| 16:55 - 17:25 | F1 | Learning Latent Action Representations in Space and Time from Demonstrations | Michael Przystupa (Vision Robotics, UofA) |
| F2 | Imitation Learning for Bimanual Insertion Tasks in Mujoco | Michael Drolet (IAS, TU Da) | |
| 17:30 - 18:00 | G1 | Slurm-tastic Adventures on a Cluster: A Roboticist Guide to Dominating the Cluster with Slurm and Docker | Daniel Palenicek (IAS, TU Da), Pascal Klink(IAS, TU Da) |
| G2 | EAGERx: Graph-Based Framework for Sim2real Robot Learning | Jelle Luijkx (CoR, TU Delft) |
Focus Groups
The aim of the focus groups is to expand participants' knowledge by delving into robotics and machine learning topics that are interesting and related but outside their primary research area. Each participant will have the opportunity to gain new insights, collaborate with peers, and spark innovative ideas. In the end, the groups will present their findings through presentations, fostering further discussion among all participants. Open to all. Chair: Ali Younes, Kay Hansel, Kai Ploeger, Snehal Jauhri
Career Seminar
The career seminar facilitates the exchange of experiences and expertise between junior and senior researchers. The seminar culminates with a Q&A session with the PIs, providing an opportunity for participants to seek specific guidance. Its overarching goal is to equip junior researchers with insights and experiences to advance their careers effectively. Open to all. Chair: Jan Peters
Grant Planing & Proposal Writing
Grant planning and proposal writing are crucial in academia as they enable researchers to secure funding for their projects, support their research efforts and expand their scientific knowledge. This seminar will specifically focus on teaching junior researchers how to plan grants and write proposals. Additionally, experiences and expertise will be exchanged between the research groups. Open to all. Chair: Jan Peters
Network and Research (N&R) speed dating
To form new collaborations, participants will be organized into small groups, intentionally aiming to have a diverse and heterogeneous mix of researchers from different labs. Within these groups, pairs of participants will engage in five-minute chats, allowing them to get to know each other. Open to all. Chair: Kay Hansel, Kai Ploeger
Poster Session
Poster session to present research, exchange ideas and get to know each other. The poster session starts before dinner and continues throughout the evening. Open to all. Chair: Kay Hansel, Kai Ploeger
Group-Specific Session
Each group is free to have its own meeting to discuss various internal matters or use these slots freely. Chair: Kay Hansel, Kai Ploeger
Travel Arrangements
Please contact Snehal Jauhri for travel advice.
Equipment
It is highly recommended to bring with you:
- Your Laptop: Please bring your laptop as it will be needed during certain sessions
- Hiking Boots: Since we'll be going for a hike or klettersteig, a sturdy pair of hiking boots is essential. Consider choosing ones that have:
- Small Backpack (20 - 30l): A convenient backpack will come in handy for carrying your essentials.
- Refillable Water Bottle: Staying hydrated is vital, so don't forget to bring a refillable water bottle.
- Small Lunch Box: We'll provide a breakfast buffet, and you can pack your lunch in a small box to keep your energy up during the event.
- Shades and Hat: The sun can be intense, so pack sunglasses and a hat to shield yourself.
- Sunscreen: To protect your skin from the mountain sun, remember to pack sunscreen.
- Rain Jacket: While we hope for clear skies, it's best to be prepared. Bring a lightweight rain jacket just in case.
- Comfortable Clothes: Expect to get a bit wet and dirty during our mountain adventures, so comfortable and durable clothes are recommended.
- Swimming Suit: There's a possibility of visiting thermal springs or a sauna, so pack a swimming suit if you'd like to take a refreshing dip.
The International Workshop on Intelligent Autonomous Learning Systems is a yearly by-invitation-only workshop on developments within machine learning for robotics and other autonomous systems. The focus lies on fundamental scientific and organizational discussions. The workshop typically takes place in Kleinwalsertal (Austria), a nice location surrounded by the beautiful Alps.
International Workshop of Intelligent Autonomous Learning Systems 2022
Quick Facts
| Organizers: | Joao Carvalho, Pascal Klink, Joe Watson |
| Location: | Darmstädter Haus, Kleinwalsertal, Austria |
| Dates: | 14.08.- 21.08.2022 |
Schedule
Note, the schedule is subject to change, to optimize the weather during outdoor activites.
Daily Schedule
| 14/8 | Sunday: | Travel. |
|---|---|---|
| 15/8 | Monday: | Introductions, Tutorial #1, Networking |
| 16/8 | Tuesday: | Talk #1, Tutorial #2 |
| 17/8 | Wednesday: | Talk #2, Group Discussions |
| 18/8 | Thursday: | Hike, Tutorial #3 |
| 19/8 | Friday: | Group discussions, Hackathon |
| 20/8 | Saturday: | Discussions / Swimming pool, Sauna |
| 21/8 | Sunday: | Travel back |
Detailed Schedule
Note, there will be 30 minute coffee breaks each morning and afternoon, between sessions.
| Time | Monday | Tuesday | Wednesday | Thursday | Friday |
|---|---|---|---|---|---|
| 09 - 10 | Introduction | Hike | Open-source Reinforcement Learning | A Robotic Platform for Air Hockey | Geri Neumann (ALR, KIT) |
| 10 - 11 | Network Speed Dating | Hike | Information-Theoretic Sensorimotor Behaviour I | Information-Theoretic Sensorimotor Behaviour II | Latent Predictive World Models |
| 11 - 12 | WIAWIWTB | Hike | Information-Theoretic Sensorimotor Behaviour I | Information-Theoretic Sensorimotor Behaviour II | Latent Predictive World Models |
| 12 - 13 | Lunch | Hike | Lunch | Lunch | Lunch |
| 13 - 14 | WIAWIWTB | Hike | (IAS) Ops, DevOps & IT Discussion | ML for Drug Discovery | Joni Pajarinen (Aalto) |
| 14 - 15 | WIAWIWTB | Hike | (IAS) Feedback Discussion | Research Speed Dating | Impactful Research Workshop |
| 15 - 16 | WIAWIWTB | Hike | (IAS) Teaching Review | (IAS) Research Review | |
| 16 - 17 | Hike | (IAS) 360° Feedback | (IAS) Research Review |
Invited Tutorials
Information-Theoretic Treatment of the Perception-Action Loop
Daniel Polani is a professor of Artificial Intelligence and Director of the Centre for Computer Science and Informatics Research (CCSIR), and Head of the Adaptive Systems Research Group, and leader of the SEPIA (Sensor Evolution, Processing, Information and Actuation) Lab at the University of Hertfordshire. Daniel Polani received the Doctor of Natural Sciences degree from the Faculty of Mathematics and Computer Science, University of Mainz, Mainz, Germany, in 1996.,From 1996 to 2000, he was a Research Assistant with the Faculty of Mathematics and Computer Science, University of Mainz. In 1997, he was a Visiting Researcher with the Neural Networks Research Group, University of Texas at Austin, Austin, TX, USA. From 2000 to 2002, he was a Research Fellow with the Institute for Neuro- and Bioinformatics, University of Lübeck, Lübeck, Germany. From 2002 to 2008, he was a Principal Lecturer with the Adaptive Systems and Algorithms Research Groups, School of Computer Science, University of Hertfordshire (UH), Hatfield, U.K. Since 2008, he has been a Reader in Artificial Life and Member of the Adaptive Systems and Algorithms Research Groups, School of Computer Science, UH. His research interests include the foundations of intelligent behavior in biological as well as artificial agents, especially in terms of informational optimality principles.
A Unified Framework for Developing Curiosity-driven Agents in Hard-Exploration Tasks using Latent-Predictive World-models. (Remote)
Mohammad Gheshlaghi Azar is a research scientist at DeepMind woking on broad range of topics from deep reinforcement learning to self-supervised learning. His main passion is to gain better understanding of the concept of intelligence by looking deeper on its governing mathematical principles.
Contributed Talks
Machine Learning for Drug Discovery
Alexander I. Cowen-Rivers is a PhD student in the Intelligent Autonomous Systems group and Research Scientist at Huawei Research London, working on Reinforcement Learning and Bayesian Optimisation. Alexander joined the lab in November 2019 after receiving his MSc in Machine Learning & Data Science from University College London in October 2018.
A Robotic Platform for Air Hockey
Puze Liu is a PhD student in the Intelligent Autonomous Systems group. His research interests include robotics, reinforcement learning, inverse reinforcement learning, and human-robot collaboration. Puze is looking for new approaches for robots to learn collaborative tasks that enable robots assisting humans in different scenarios.
Open-source Reinforcement Learning
Davide Tateo is a postdoctoral researcher in the Intelligent Autonomous Systems group working on Robotics and Reinforcement Learning.
Workshop Activities
WIAWIWTB (Where I Am & Where I Want To Be)
A series of brief 3 minute lightning talks summarizing your research career so far and future aspirations for all retreat participlants. Talks are up to three minutes long, with two minutes allocated per speaker for discussion. Chair: Joe Watson
Impactful Research Workshop
What makes impactful research? How should we choose problems to work on? This workshop aims to develop strategies for finding the good research questions and classifying dead ends, and discuss the current open questions in Robot Learning. Chair: Joe Watson
Network and Research Speed Dating
Rapid three minutes chats to get to know each other, and establish potential research collaborations. Open to all. Chair: Joe Watson
IAS Workshops
IAS-only meetings to discuss various internal matters. Other participants are free to have their own meetings or use these slots freely. Chairs: Joe Watson and Joao Carvalho
Tooling Tutorials
Contributed workshops looking at a specific problems or software libraries. Chairs: Boris Belousov (Isaac Gym), Daniel Palenicek (JAX) and Joe Watson (RevealJS)
Travel Arrangements
Please contact Pascal Klink for travel advice.