Stephen James (@stepjamUK) / X

Stephen James

513 posts

Stephen James

@stepjamUK

CEO @Neuracore_AI | Assistant Professor @imperialcollege | ex-Director of Dyson Robot Learning Lab | Postdoc @UCBerkeley w/ @pabbeel | PhD ICL w/ @ajdDavison

London, England

Joined January 2010

Pinned
Stephen James
@stepjamUK
Sep 24, 2025
𝗔𝗳𝘁𝗲𝗿 𝟭𝟬+ 𝘆𝗲𝗮𝗿𝘀 𝗶𝗻 𝗿𝗼𝗯𝗼𝘁 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴, from my PhD at Imperial to Berkeley to building the Dyson Robot Learning Lab, one frustration kept hitting me: 𝗪𝗵𝘆 𝗱𝗼 𝗜 𝗵𝗮𝘃𝗲 𝘁𝗼 𝗿𝗲𝗯𝘂𝗶𝗹𝗱 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗼𝘃𝗲𝗿 𝗮𝗻𝗱
00:00
57K
Stephen James
@stepjamUK
Oct 14, 2025
𝗗𝗟𝗥 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵𝗲𝗿𝘀 𝗴𝗮𝘃𝗲 𝗮 𝗿𝗼𝗯𝗼𝘁𝗶𝗰 𝗮𝗿𝗺 𝗳𝘂𝗹𝗹-𝗯𝗼𝗱𝘆 𝘁𝗼𝘂𝗰𝗵 𝘀𝗲𝗻𝘀𝗶𝘁𝗶𝘃𝗶𝘁𝘆 𝘄𝗶𝘁𝗵 𝗻𝗼 𝗮𝗿𝘁𝗶𝗳𝗶𝗰𝗶𝗮𝗹 𝘀𝗸𝗶𝗻 𝗻𝗲𝗲𝗱𝗲𝗱. They used internal force-torque sensors at 8 kHz + deep learning. The robot can feel where you touch it,
00:00
174K
Stephen James
@stepjamUK
Aug 18, 2025
𝗜'𝘃𝗲 𝗵𝗲𝗮𝗿𝗱 𝘁𝗵𝗶𝘀 𝗮 𝗹𝗼𝘁 𝗿𝗲𝗰𝗲𝗻𝘁𝗹𝘆: "𝗪𝗲 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝘂𝗿 𝗿𝗼𝗯𝗼𝘁 𝗼𝗻 𝗼𝗻𝗲 𝗼𝗯𝗷𝗲𝗰𝘁 𝗮𝗻𝗱 𝗶𝘁 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗲𝗱 𝘁𝗼 𝗮 𝗻𝗼𝘃𝗲𝗹 𝗼𝗯𝗷𝗲𝗰𝘁 - 𝘁𝗵𝗲𝘀𝗲 𝗻𝗲𝘄 𝗩𝗟𝗔 𝗺𝗼𝗱𝗲𝗹𝘀 𝗮𝗿𝗲 𝗰𝗿𝗮𝘇𝘆!" Let's talk about what's actually
00:00
51K
Stephen James
@stepjamUK
Jan 22, 2024
Dyson Robot Learning Lab is Hiring full-timers and interns! 🤖 1x Research Scientist: forms.gle/Z8pBnMMZpBAw8t… 1x Data Engineer (Data Collection & ML Training): forms.gle/egqd4UdQH5vV4z… 3x PhD Internship: forms.gle/4rpUY8izJUvfoT… Come join our lab of 12; located in London, UK 🇬🇧
00:00
77K
Stephen James
@stepjamUK
Oct 1, 2025
𝗧𝗵𝗲 𝗽𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝗼𝗻: Morgan Stanley forecasts a $5 trillion humanoid robot market by 2050. 𝗧𝗵𝗲 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲: Getting there will be harder than most people realise. Let me share some perspective on what it actually takes: 𝗖𝘂𝗿𝗿𝗲𝗻𝘁 𝗽𝗿𝗼𝗴𝗿𝗲𝘀𝘀 (2025):
00:00
51K
Stephen James
@stepjamUK
Jun 14, 2023
We have a Fall PhD Internship opening at the Dyson Robot Learning Lab in London! 🇬🇧 Come and join our world-class team of robot learning gurus for 3-6 months! Apply via form: forms.gle/crwXtPJoS9Xahb…
00:00
37K
Stephen James
@stepjamUK
Feb 9, 2022
New paper! Not all reinforcement learning problems are suited for a Gaussian policy parameterization!🤯 Plan to use 3D rotation or 6D pose as part of an action space? Consider the Bingham distribution! 🧵1/5 Paper: arxiv.org/abs/2202.03957 Code: github.com/stepjam/BPP w/@pabbeel
Stephen James
@stepjamUK
Oct 29, 2025
𝗛𝗲𝗿𝗲’𝘀 𝗮 𝗺𝗶𝘀𝘁𝗮𝗸𝗲 𝗜 𝘀𝗲𝗲 𝗮𝗹𝗹 𝘁𝗵𝗲 𝘁𝗶𝗺𝗲. Teams collect robot data at 30Hz because “that’s what our robot runs at” and then wonder why their models underperform. The truth is that 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝘁𝗮𝘀𝗸𝘀 𝗻𝗲𝗲𝗱 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝘁𝗲𝗺𝗽𝗼𝗿𝗮𝗹
00:00
23K
Stephen James
@stepjamUK
Aug 13, 2025
𝗣𝗼𝗽𝘂𝗹𝗮𝗿 𝗼𝗽𝗶𝗻𝗶𝗼𝗻: "𝗪𝗲 𝗻𝗲𝗲𝗱 𝗺𝗼𝗿𝗲 𝗱𝗮𝘁𝗮." 𝗔𝗰𝘁𝘂𝗮𝗹 𝗿𝗲𝗮𝗹𝗶𝘁𝘆: 𝟭𝟬𝟬 𝗴𝗿𝗲𝗮𝘁 𝗱𝗲𝗺𝗼𝘀 > 𝟭𝟬,𝟬𝟬𝟬 𝗺𝗲𝗱𝗶𝗼𝗰𝗿𝗲 𝗼𝗻𝗲𝘀. 𝗧𝗵𝗲 𝗺𝘆𝘁𝗵: More demonstrations always mean better models. 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹𝗶𝘁𝘆: I've seen models trained
00:00
35K
Stephen James
@stepjamUK
Sep 23, 2025
𝗘𝘃𝗲𝗿𝘆𝗼𝗻𝗲’𝘀 𝘁𝗮𝗹𝗸𝗶𝗻𝗴 𝗮𝗯𝗼𝘂𝘁 “𝗣𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗔𝗜" - the idea that we can simulate real-world environments so well that robots trained in simulation will work perfectly in reality. 𝗧𝗵𝗲 𝗽𝗿𝗼𝗺𝗶𝘀𝗲: Train in virtual worlds → deploy anywhere. 𝗧𝗵𝗲
00:00
25K
Stephen James
@stepjamUK
Mar 30, 2022
Following the success of Masked AutoEncoders (MAE), we all knew it was coming.... VideoMAE. The death of contrastive learning for video representation learning? I think so. Props to the rapid pace of the authors.👏 MAE has only been out for ~3 months🤯 arxiv.org/pdf/2203.12602…
Stephen James
@stepjamUK
Jun 8, 2022
A year ago, I thought it obvious that meta-RL had a role to play in robotics... Now I’m no longer convinced! Our large-scale study shows multi-task pretraining followed by fine-tuning on novel tasks, performs >= meta-RL! Lead: @ZhaoMandi w/@pabbeel arxiv.org/abs/2206.03271
Stephen James
@stepjamUK
Dec 20, 2018
Sim-to-Real via Sim-to-Sim: we learn a generator that translates real-world images to a canonical simulation version to learn robot grasping with no real-world data! arxiv.org/abs/1812.07252 w/ collaborators from @GoogleAI, @Theteamatx, @DeepMindAI
GIF
Stephen James
@stepjamUK
Oct 2, 2025
𝗡𝗶𝗻𝗲 𝘆𝗲𝗮𝗿𝘀 𝗮𝗴𝗼, 𝗜 𝘀𝘁𝗼𝗼𝗱 𝗶𝗻 𝗮 𝘀𝗺𝗮𝗹𝗹 𝗚𝗼𝗼𝗴𝗹𝗲 𝗰𝗼𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝗿𝗼𝗼𝗺 𝘄𝗶𝘁𝗵 𝟭𝟬𝟬 𝗽𝗲𝗼𝗽𝗹𝗲, 𝘀𝗵𝗮𝗿𝗶𝗻𝗴 𝗺𝘆 𝗳𝗶𝗿𝘀𝘁 𝗣𝗵𝗗 𝘄𝗼𝗿𝗸. 𝗧𝗼𝗱𝗮𝘆, 𝘁𝗵𝗮𝘁 𝘀𝗺𝗮𝗹𝗹 𝗿𝗼𝗼𝗺 𝗵𝗮𝘀 𝗯𝗲𝗰𝗼𝗺𝗲 𝗮 $𝟭𝟲.𝟱𝗕 𝗶𝗻𝗱𝘂𝘀𝘁𝗿𝘆...
21K