Andrew Davison
@AjdDavison
From SLAM to Spatial AI; Professor of Robot Vision, Imperial College London; Director of the Dyson Robotics Lab; Co-Founder of Slamcore. FREng, FRS.
At #ICML2024 today Seth presents GBP Learning *within* factor graphs. Instead of weights and neurons, we use variables and non-linear factors, and optimise via Gaussian Belief Propagation, unifying continual learning and inference and allowing integration with state estimation.
Excited to be at ICML! I'll be presenting a poster on GBP Learning — our approach to train deep networks using Gaussian belief propagation. With @markvanderwilk and @AjdDavison. Poster #1606 between 11:30-1pm today.
Excellent job opportunity for a SLAM and Spatial AI Research Engineer in the growing team at Slamcore in London.
Lead Research Engineer (SLAM/ State estimation) Build SLAM & spatial AI deployed in real environments. 🌟 We're hiring! 🌟 You'll design localisation & mapping algorithms, deploy real-world robotics solutions, and work with a top-tier team. 👉 Strong CV/SLAM/C++ skills 🏠…
I'm hearing the term Spatial AI all the time now! Various exciting new companies and projects popping up. #SpatialAI
6 years on from when (I believe) the term was coined in this paper, and also used by @SlamcoreLtd, some big companies I know that now have named Spatial AI teams: Meta (led by Jakob Engel) Microsoft (led by Marc Pollefeys) Naver Labs (led by Christian Wolf) #SpatialAI
Excellent opportunity to join Slamcore in London. #SpatialAI
Lead Research Engineer (SLAM/ State estimation) Build SLAM & spatial AI deployed in real environments. 🌟 We're hiring! 🌟 You'll design localisation & mapping algorithms, deploy real-world robotics solutions, and work with a top-tier team. 👉 Strong CV/SLAM/C++ skills 🏠…
Kasparov of our times
Humanity has prevailed (for now!) I'm completely exhausted. I figured, I had 10h of sleep in the last 3 days and I'm barely alive. I'll post more about the contest when I get some rest. (To be clear, those are provisional results, but my lead should be big enough)
This is a great opportunity. I am collaborating with Danyal on work closely related to this and he is full of great ideas.
I'm hiring a @ARIA_Research funded position at @imperialcollege @ImperialEEE. Broad in scope + flexible on topics! Exciting projects on intersection of neural networks & new AI accelerators, from a HW/SW co-design perspective. Join us 🚀: imperial.ac.uk/jobs/search-jo…
I still miss Network Wars. @jim_rutt will it be coming back to Android?
Network Wars is my favourite mobile game by miles networkwars.com but I just changed to a new phone and was disappointed to see it doesn't seem to be on the Google Play Store any more. @jim_rutt will it be back?
Reminds me of A Difficult Game About Climbing share.google/F9sgDSsNefWPHr…
HUMAN vs @MyolabAI's athletic intelligence Human athletic intelligence — captured and analyzed by MyoLab.AI Not just another LLM, We’re building 𝑴𝒚𝒐𝑺𝒂𝒑𝒊𝒆𝒏𝒔 — the world’s first Human Embodied Intelligence.
Happy to be at ICVSS in Sicily this week. The theme this year is #SpatialAI!
Great talk by Andrew Davison on the journey from SLAM to Spatial AI at #ICVSS2025! A compelling argument for the power of dynamic, updatable 3D representations and their direct impact on the future of robotics and wearable devices. And nice insights for spatial AI researchers!
fwiw, I think Prof. @percyliang and the CS336 team nailed this: Sutton’s Bitter Lesson is often misinterpreted as “scale is all that matters” and/or “algorithms don’t matter.” The more accurate – and useful – interpretation is: what matters are the algorithms that scale.…
(1) a big part of being human is language - a library of "software programs" that help us live in the world (2) a billion neurons in a bird brain is enough to create a smart, generalized, agentic intelligence we've figured out (1) but have barely scratched the surface for (2)
Went to see this yesterday, recommended!
To celebrate International Year of Quantum Science & Technology the @nqcc_official has convened partners from across the UK to build an exhibit for the @royalsociety’s Summer Science Exhibition. Find us at the Quantum Zone until 6 July! Read more nqcc.ac.uk/summer-science… #IYQ2025
Well done to Riku and @eric_dexheimer on Best Demo Honourable Mention for MASt3R-SLAM at #CVPR2025. That's 3 best demo awards in the past 2 years at CVPR from our lab! (Gaussian Splatting SLAM won Best Demo in 2024 and SuperPrimitive got Honourable Mention). #DemosNotTables
We (with @eric_dexheimer) will be demoing MASt3R-SLAM @CVPR today 16:00-18:00, ExHall B, 9 Come and see the limits of our real-time monocular dense SLAM! We’ll also be presenting our poster #83 on Sat 14th from 17:00-19:00. See you soon!
Congratulations @jianyuan_wang, and well said.
Just touched down from the air back to solid ground😄 We're honored to receive the Best Paper Award this year, while I want to give a special shout-out to the authors of the brilliant paper Dust3R: Shuzhe Wang (@riverakid1), Vincent Leroy (@Vinc3nt_Leroy), Yohann Cabon, Boris…
Live Demo of MASt3R-SLAM happening this afternoon in the demo session at #CVPR2025.
We’ve had fun testing the limits of MASt3R-SLAM on in-the-wild videos. Here’s the drone video of a Minnesota bowling alley that we’ve always wanted to reconstruct! Different scene scales, dynamic objects, specular surfaces, and fast motion.
Mostly agree with this.
I glimpsed during my work in Ian Witten‘s lab in the 1990ies that one can extract arbitrary semantics from pure statistical text models. I just never considered this to be very interesting, because I thought AGI would result from intelligent self improvement, not mere imitation.
Congrats to Wayve on this. I got to visit and experience an autonomous ride in London a few weeks ago and was extremely impressed.
Big news: @wayve_ai is partnering with @Uber @dkhos @andrewgordonmac to launch the first-ever public road trials of Level 4 fully driverless vehicles in London 🇬🇧🚗 London is our home and one of the most diverse cities - where I can't wait to bring the benefits of autonomy!
Prediction: eventually, in AI, all graphical models (call them networks or AI models if you want) will be factor graphs. Undirected graphs with locally-defined properties that enable modular, fully distributed, and even heterogeneous and asynchronous computation.
All researchers should fight against this. Every week I try to persuade my students that top papers often have few quantitative results. With work that's new, important, and clearly qualitatively different (zero to one!), you don't need quantitative results. Demos not tables!
The author's dilemma, circa 2021
Great to see @imperialcollege alumni from our lab @czarnowskij and Tristan leading this brilliant work on practical humanoid manipulation.
In their latest video, Boston Dynamics’s AI team explains how they make the Atlas humanoid perceive and interact with the world. Atlas uses an agile perception system to understand both the shape and context of objects in complex environments. Atlas combines 2D and 3D…