Workshop on World Models and Predictive Coding in Cognitive Robotics
@IEEE IROS2023

The workshop has been accepted to IROS 2023.
This workshop will  be held in conjunction wiht the 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2023) will be held October 5, 2023 at Huntington Place in Detroit, Michigan, USA. 

Abstract

This workshop will explore the new frontiers in robotics, emphasizing world models, predictive coding, probabilistic generative models, and the free-energy principle. Creating autonomous robots that can actively explore the environment, acquire knowledge and learn skills continuously is the ultimate achievement envisioned in cognitive and developmental robotics. Importantly, if the aim is to create robots that can continuously develop through interactions with their environment, their learning processes should take into account the interactions with their physical and social world in the manner of human learning and cognitive development. Based on this context, this workshop focuses on the two concepts of world models and predictive coding in the context of cognitive robotics. Recently, world models have attracted renewed attention as a topic of considerable interest in artificial intelligence. Cognitive systems learn world models to better predict future sensory observations and optimize their policies, i.e., controllers. Alternatively, in neuroscience, predictive coding proposes that the brain continuously predicts its inputs and adapts to model its own dynamics and control behavior in its environment. Both ideas may be considered as underpinning the cognitive development of robots and humans capable of continual or lifelong learning. The workshop aims to provide a platform for researchers and practitioners in cognitive, developmental and bioinspired robotics, and computational and theoretical neuroscience to exchange ideas and explore new directions. This workshop will provide a valuable experience for all the participants and will significantly contribute to the growth and development of the field of cognitive robotics.

Content of the workshop

Creating autonomous robots that can actively explore the environment, acquire knowledge and learn skills continuously is the ultimate achievement envisioned in cognitive and developmental robotics. Importantly, if the aim is to create robots that can continuously develop through interactions with their environment, their learning processes should take into account the interactions with their physical and social world in the manner of human learning and cognitive development. Based on this context, this workshop focuses on the two concepts of world models and predictive coding in the context of cognitive robotics. Recently, world models have attracted renewed attention as a topic of considerable interest in artificial intelligence. Cognitive systems learn world models to better predict future sensory observations and optimize their policies, i.e., controllers. Alternatively, in neuroscience, predictive coding proposes that the brain continuously predicts its inputs and adapts to model its own dynamics and control behavior in its environment. Both ideas may be considered as underpinning the cognitive development of robots and humans capable of continual or lifelong learning.

This workshop will explore the new frontiers in robotics, emphasizing world models, predictive coding, probabilistic generative models, and the free-energy principle. 

The workshop aims to provide a platform for researchers and practitioners in cognitive, developmental and bioinspired robotics, and computational and theoretical neuroscience to exchange ideas and explore new directions. The workshop will reflect the state of the art on the topic by focusing on the latest research trends and advancements in world models and predictive coding in cognitive robotics. The keynote speakers will provide insights into the most recent findings in these areas. Importantly, many of the organizers have conducted a survey and written a review paper called “World Models and Predictive Coding for Cognitive and Developmental Robotics: Frontiers and Challenges” recently [1]. The keynote speakers will be invited based on the survey. This will ensure that the workshop will become a cutting-edge event. We will also encourage participants to submit their research papers, which will go through a rigorous review process to ensure that they reflect the current state of the art.

We will engage participants and exchange ideas through various activities such as interactive sessions, group discussions, and poster presentations. The workshop will provide an opportunity for participants to network, collaborate, and learn from each other. We will also organize a panel discussion to encourage an open and honest discussion about the challenges and opportunities of world models and predictive coding in cognitive robotics.

The event will expand the content diversity of IROS 2023 by providing a platform for researchers and practitioners in cognitive robotics to present and discuss their research in world models, predictive coding, probabilistic generative models, and the free energy principle. The workshop also provides a unique opportunity for interaction between developmental robotics, supported by the IEEE CIS TC Cognitive Developmental Systems, and cognitive robotics, supported by the IEEE RAS TC Cognitive Robotics. The collaboration of two very related TCs is expected.

In summary, this workshop will provide a unique opportunity for researchers and practitioners in cognitive robotics to explore the latest advances in world models and predictive coding. Through this workshop, we aim to promote the development of autonomous cognitive and developmental robots and encourage the use of world models and predictive coding in cognitive robotics. We believe that this workshop will be an enriching experience for all participants and will contribute to the growth and development of the field of cognitive robotics.

[1] Taniguchi, T., Murata, S., Suzuki, M., Ognibene, D., Lanillos, P., Uğur, E., Jamone, L., Nakamura, T., Ciria, A., Lara, B., & Pezzulo, G. (2023). World Models and Predictive Coding for Cognitive and Developmental Robotics: Frontiers and Challenges. Advanced Robotics, 37:13, 780-806, DOI: 10.1080/01691864.2023.2225232  (ArXiv, abs/2301.05832. https://arxiv.org/abs/2301.05832)

Program

Session 1 (Chair: Tadahiro Taniguchi)

8:30 - 8:40

Welcome and Introduction [Slide]

Tadahiro Taniguchi (Ritsumeikan University)

8:40 - 9:25

Invited Talk 1 (from Organizers)

Masahiro Suzuki (University of Tokyo, Japan)

"Perspectives on World Models and Predictive Codings in Cognitive Robotics" 

9:25 - 10:10

Invited Talk 2

Tetsuya Ogata (Waseda University, Japan) 

"Predictive Coding-inspired Robotics: Advancing Adaptability through Deep Predictive Learning"

10:10 – 11:00

Coffee break

Session 2 (Chair: Tadahiro Taniguchi)

11:00 - 11:50

Flash poster and spotlight session

• Oral  (15 min. talk + 5 min. QA)

  1. World-Model-Based Control for Industrial box-packing of Multiple Objects using NewtonianVAE
     Yusuke Kato, Ryo Okumura, Tadahiro Taniguchi

• Spot light (5 min. talk + 3 min. QA)

  1. Memory Development with Heteroskedastic Bayesian Last Layer Probabilistic Deep Neural Networks
     Georgios Velentzas, Costas Tzafestas, Mehdi Khamassi

  2. Learning to Navigate from Scratch using World Models and Curiosity: the Good, the Bad, and the Ugly
     Daria de Tinguy, Sven Remmery, Pietro Mazzaglia, Tim Verbelen and Bart Dhoedt

  3. Tactile In-Hand Pose Estimation through Perceptual Inference
     Tatsuya Kamijo, Tomoshi Iiyama, Yuta Oshima, Gentiane Venture, Tatsuya Matsushima, Yutaka Matsuo and Yusuke Iwasawa

• Flash talks (1 min. talk)

  1. Video Understanding through Programs of Action
     Eadom Dessalene, Michael Maynord, Cornelia Fermu¨ller, Yiannis Aloimonos

  2. Conditional Neural Processes for Self-Supervised Goal Discovery in Action Imitation
     Marco Gabriele Fedozzi, Yukie Nagai, Francesco Rea, Alessandra Sciutti

  3. Learning Spatial and Temporal Hierarchies: Hierarchical Active Inference for navigation in Multi-Room Maze Environments
     Daria de Tinguy, Toon Van de Maele, Tim Verbelen and Bart Dhoedt 

11:55 - 12:30

Poster session 1

12:30 – 13:30

Lunch

Session 3 (Chair: Yukie Nagai)

13:30 - 14:15

Invited Talk 3

Giulio Sandini (Italian Institute of Technology, Italy)

“The Role of Imagination in Social Interaction”

14:15 - 15:00

Invited Talk 4

Pablo Lanillos (Donders Institute for Brain, Cognition and Behaviour, Netherlands)

"Neuroscience-inspired generative models of action"

15:00 – 16:00

Coffee break (with poster session 2)

Session 4 (Chair: Tetsunari Inamura)

16:00 - 16:45

Invited Talk 5

Takamitsu Matsubara (NAIST, Japan)

"Disturbance-injected Robust Imitation Learning with Structured Policies for Complex Tasks"

16:45 - 17:15

Open & Panel discussion

Modelator: Tetsunari Inamura (Chair of TC Cognitive Robotics)

17:15 - 17:30

Closing remarks
(including Award Ceremony)

17:30

End

Organizers

1. Tadahiro Taniguchi, Ritsumeikan University,

2. Emre Ugur, Bogazici University,

3. Masahiro Suzuki, The University of Tokyo,

4. Dimitri Ognibene, Università degli Studi di Milano-Bicocca,

5. Lorenzo Jamone, Queen Mary University of London,

6. Yukie Nagai, The University of Tokyo, 

7. Tatsuya Matsushima, The University of Tokyo,

8. Tetsunari Inamura, Tamagawa University,

Website

https://world-model.emergent-symbol.systems/home

Sponsor

• IEEE Technucal Committee for Cognitive Robotics

• JST Moonshot Project Goal 3 “Realization of AI robots that autonomously learn, adapt to their environment, evolve in intelligence and act alongside human beings, by 2050

Endorcement

• IEEE RAS Technucal Committee for Cognitive Robotics

• IEEE CIS Technical Committee Cognitive Developmental Systems