🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Super excited to share our new work - a personal first - Variational Causal Dynamics (VCD). We trained a latent state space model with a causal and modular structure for efficient adaptation. Work at Oxford Applied AI Lab with Bernhard Schölkopf, Ingmar Posner Paper: arxiv.org/abs/2206.11131 1/3

Admittedly, I've been a doubter when it comes to all things causal. That's why we set ourselves the challenge of learning a modular world model drawing on causal discovery. Very excited about where this is going, Anson Lei, Bernhard Schölkopf. Oxford Applied AI Lab Intelligent Systems, Oxford Robotics Institute

Excited to present AMP-LS, our recent work on gradient-based motion planning, accepted to IEEE ICRA 2023. AMP-LS can plan a collision-free trajectory efficiently in novel complex scenes and is trained on only easily generated kinematically valid joint states. 🧵👇

Our next offering in planning for manipulation using optimisation in structured latent spaces: fast, reactive planning behaviour in the real world. After best poster at the 4th UK Manipulation Workshop, come and see us IEEE ICRA 2023! Oxford Applied AI Lab Oxford Robotics Institute

🧵 Introducing RAMP, an open-source robotics benchmark inspired by real-world industrial assembly tasks. Check out the RAMP benchmark website: sites.google.com/oxfordrobotics…. 1/5 👇 w/ M. Robson Jun Yamada M. Sridharan K. Janik Ingmar Posner Oxford Applied AI Lab Oxford Robotics Institute The MTC

The intricacies of latent-space planning for robust locomotion are explored in a newly published T-RO paper by researchers Oxford Applied AI Lab and Oxford Dynamic Robot Systems. ieeexplore.ieee.org/document/10197… #LeggedLocomotion #QuadrupedalRobots #RobotLearning #AE

Autoregressive next-token prediction is not enough: reliable AI agents are going to require accurate models of the world. I’m excited to share a new approach to world modeling that does not require autoregressive sampling: “World Models via Policy-Guided Trajectory Diffusion”…

Can we generate on-policy trajectories using a "non-autoregressive" world model to alleviate compounding errors? — Yes! We introduce "PolyGRAD", a non-autoregressive diffusion world model with policy guidance to generate accurate on-policy trajectories. led by Marc Rigter

Inspired work introducing policy-guided #diffusion by Marc Rigter & Jun Yamada . The prospect of efficiently imagining entire on-policy trajectories in one go is tantalising. Looking forward to exploring where this can take us... Oxford Applied AI Lab Oxford Robotics Institute Engineering Science, Oxford

Interested in efficient model-based RL in the real world using visual observations? Then here is one more worth checking out this year: World-Model Distillation, lead by Jun Yamada (together with Marc Rigter and Jack Collins) ... Oxford Applied AI Lab #RobotLearning #Robotics

If you are excited about world-models in robotics, check out this EPSRC iCASE PhD studentship (fully funded for UK Home students) in Oxford Applied AI Lab in collaboration with Siemens: Foundation Models for Industrial Control Applications. *Deadline: 1 March 2024* ox.ac.uk/admissions/gra…

We introduce D-Cubed, a novel trajectory optimisation method using a latent diffusion model trained from a task-agnostic play dataset, including only representative hand motions, to solve dexterous deformable object manipulation tasks! (1/N)

Trajectory optimisation in high dimensional spaces is notoriously hard. What if you could leverage basic experience of what the system can do and let a diffusion model and vanilla sim guide you? Stunning work led by Jun Yamada with Shaohong Zhong and Jack Collins Oxford Applied AI Lab

Delighted to be at #ICRA2024. Interested in effective sim-2-real transfer for world models (WeBT7-CC.6)? Or benchmarking for robot assembly (ThAT9-CC.3)? Or predicting lane graphs for autonomous driving (ThBT6-CC.2)? Come and see us to meet, discuss, or just hang-out...

🚀 We’re hiring a Postdoc in Multimodal World Modelling for robot skill acquisition! 🌟 Do you have a passion for deploying AI on real-world robots? Then this one may be for you... my.corehr.com/pls/uoxrecruit… #robotlearning #Robotics #GenerativeAI #MachineLearning #AI Oxford Robotics Institute

Very excited to share our new work - SPARTAN: A Sparse Transformer Learning Local Causation. We develop a Transformer world model that learns local causal dependencies between entities, leading to improved adaptation efficiency and robustness with accurate prediction. 🧵

How can a transformer uncover local causal dependencies in dynamic systems, from simulations to real-world data? 🤔 The answer: Hard attention + sparsity. But with a twist. Meet SPARTAN: More causal. More efficient. Just as accurate. #Robotics #ML #AI #CausalAI

🚀 Funded PhD Studentship Alert! 🚀 Join Oxford Applied AI Lab in our mission to advance World Models for robotics and beyond. Two opportunities starting Oct 2025. Deadlines in January 2025. Details in 🧵… Oxford Robotics Institute 🤖 #Robotics #ArtificialInteligence #MachineLearning

I’m pleased to announce our work which studies complexity phase transitions in neural networks! We track the Kolmogorov complexity of networks as they “grok”, and find a characteristic rise and fall of complexity, corresponding to memorization followed by generalization. 🧵

Amongst my favourite research directions this year: understanding model complexity and its link to generalization and intelligence. Progress here could mean leaner models, versatile representations, and less reliance on data/energy. Excited that we’re off to the races on this!