Luca Rigazio

while(1) { think(); do(); }

[Luca Rigazio @gigazio]

Proud of my team @_italai joining the pinnacle of robotics manipulation research with @berkeley_ai

[Dantong Niu @Dantong_Niu]

Excited to share T-Rex: Tactile-Reactive Dexterous Manipulation 🦖🤖

Touch is fundamental to human dexterity, yet most Vision-Language-Action (VLA) models either ignore tactile feedback or lack the ability to react to high-frequency contact signals.

In this work, we tackle both the data and architectural challenges of tactile-reactive dexterous manipulation.

🦖 A 100-hour tactile-synchronized dexterous manipulation dataset with 7,700+ trajectories, 22 motor primitives, and 200+ everyday objects.

🦖 A tactile-reactive MoT architecture with spatial-temporal tactile encoding and asynchronous high-frequency tactile refinement.

🦖 A scalable training recipe combining 22,889 hours of human egocentric pretraining with tactile-grounded robot mid-training.

Across 12 real-world contact-rich manipulation tasks, T-Rex achieves over 30% higher average success rate than the strongest baseline.

We are fully open-sourcing the dataset, models, teleoperation stack, training code, and inference pipeline.

🌐 Project: https://t.co/AiHKRR8YXU
📄 Paper: https://t.co/mXY2UNLlqc
💻 Code: https://t.co/7skCxUtwKC
🤗 Dataset: https://t.co/uNwW8dcRZL
🧵 Thread ↓

[Dantong Niu @Dantong_Niu]

Excited to share T-Rex: Tactile-Reactive Dexterous Manipulation 🦖🤖

Touch is fundamental to human dexterity, yet most Vision-Language-Action (VLA) models either ignore tactile feedback or lack the ability to react to high-frequency contact signals.

In this work, we tackle both the data and architectural challenges of tactile-reactive dexterous manipulation.

🦖 A 100-hour tactile-synchronized dexterous manipulation dataset with 7,700+ trajectories, 22 motor primitives, and 200+ everyday objects.

🦖 A tactile-reactive MoT architecture with spatial-temporal tactile encoding and asynchronous high-frequency tactile refinement.

🦖 A scalable training recipe combining 22,889 hours of human egocentric pretraining with tactile-grounded robot mid-training.

Across 12 real-world contact-rich manipulation tasks, T-Rex achieves over 30% higher average success rate than the strongest baseline.

We are fully open-sourcing the dataset, models, teleoperation stack, training code, and inference pipeline.

🌐 Project: https://t.co/AiHKRR8YXU
📄 Paper: https://t.co/mXY2UNLlqc
💻 Code: https://t.co/7skCxUtwKC
🤗 Dataset: https://t.co/uNwW8dcRZL
🧵 Thread ↓

[Luca Rigazio @gigazio]

Attended the MJ Med event in SF - @DavidSHolz back building hardware is great news for humanity!

[Midjourney @midjourney]

Announcing a new division of Midjourney called "Midjourney Medical"

[Midjourney @midjourney]

Announcing a new division of Midjourney called "Midjourney Medical"

[Luca Rigazio @gigazio]

As an historic reference this is what happens when high capacity deep models with the right architecture hits a domain where data is available and legacy methods were stuck for decades @CVPR KeyNote

[Luca Rigazio @gigazio]

@CVPR KeyNote on Programmable Biology - we will cure all diseases in a decade or two and then we will cure death

[Luca Rigazio @gigazio]

@DJ_CURFEW If your best engineers are spending time reviewing other people's code, then this is inherently an inefficient bottleneck. These engineers can review their agent's code much faster than reviewing human code.

==> why?

[Luca Rigazio @gigazio]

@garrytan Not the west. America.
Italy food industry is still primarily family run. So it’s Spain and to some extent Italy - likely Greece.
No wonder the service and quality, and also the prices, it’s better in all directions all at once (not a compromise).

[Luca Rigazio @gigazio]

@lifeonskis_ I think it’s called Tignes not Tignes …

[Luca Rigazio @gigazio]

@ThoNg676733 It’s because it’s the best song to dance to

[Luca Rigazio @gigazio]

Why? Did your robo surgery went badly?

[Bojan Tunguz @tunguz]

Starting to suspect that he might not be the genius that everyone thinks he is.

[Luca Rigazio @gigazio]

@ns123abc They just lost all cofounders. Nobody care about this they are fourth in the US over 4 contenders and likely behind a bunch of Chinese and French teams that have 1/100 capital. Nobody cares

[Luca Rigazio @gigazio]

@buccocapital Agreed

[Luca Rigazio @gigazio]

I concur

[Luca Rigazio @gigazio]

@ItaliAuto Stelvio! Here is us with “Le Bolide”!

[Luca Rigazio @gigazio]

@earthcurated Ken il Guerriero!

[Luca Rigazio @gigazio]

Damn. Neurips is a behemoth

[Luca Rigazio @gigazio]

@Skiminok I wonder what’s up with flights to San Diego these days 😉

[Luca Rigazio @gigazio]

@otokyo 27

[Luca Rigazio @gigazio]

@shaneguML Definitely