Gabriele Tiboni

Co-Founder & Tech @ mkind
Building intelligent robotics for high-mix manufacturing tasks.

[Rickmer Krohn @RKrohn_]

Solving contact-rich manipulation tasks even under disco light🪩?
Incorporate multiple sensors using MSDP and obtain a robust policy in under 55 min!
Excited to share our work accepted at RA-L: Self-Supervised Multisensory Pretraining for Contact-Rich Robot Reinforcement Learning

[Gabriele Tiboni @GabrieleTiboni1]

You can join at: https://t.co/Lw3e6xwlTO

[Gabriele Tiboni @GabrieleTiboni1]

Did you know Paperedge (.ai) organically reached early adopters from 18 different countries and almost 50 different affiliations?

👉 Today we're releasing a new batch of *50 early access slots* via the link in 1st comment below 👈

On a quest to make research less overwhelming.

[Gabriele Tiboni @GabrieleTiboni1]

@paperedgeapp Things we might add in the future:
- Frequently asked LLM queries by other users
- Collaborative LLM chatting with other users
- "View as a Blogpost" to turn the list of papers into a full blogpost on the fly — with text, tables and images drawn from the respective papers

[Gabriele Tiboni @GabrieleTiboni1]

One feature I love about @paperedgeapp is "public collections":

💡 I created a public collection on "Sim-to-Real for Robotics" with 23 seed papers [see link in comment below].
🌐 Anyone can access it, add new papers and chat with the papers in context.

[Gabriele Tiboni @GabrieleTiboni1]

@paperedgeapp For some reason the community hasn't done a great job in finding ways to keep well-organized and collaborative lists of papers per topic.
I think we can do better than static survey papers and awesome-* github repos.

[Gabriele Tiboni @GabrieleTiboni1]

@paperedgeapp What even is Paperedge? You can learn more and get early access via my previous post on X: https://t.co/O7eW7UXvui

[Gabriele Tiboni @GabrieleTiboni1]

We've quietly rebuilt a Zotero-like tool over the past 6 months (@paperedgeapp):
- Free, forever
- AI chat with your entire library in context.
- Collaborative by design
Today we're opening 100 early adopter slots: first-come-first-serve via the link below🙂

[Gabriele Tiboni @GabrieleTiboni1]

@paperedgeapp Public collection on "Sim-to-Real for Robotics": https://t.co/SwMNcltOwQ

[Gabriele Tiboni @GabrieleTiboni1]

Get your early access spot at: https://t.co/5vrmaiytUy

[Gabriele Tiboni @GabrieleTiboni1]

We've quietly rebuilt a Zotero-like tool over the past 6 months (@paperedgeapp):
- Free, forever
- AI chat with your entire library in context.
- Collaborative by design
Today we're opening 100 early adopter slots: first-come-first-serve via the link below🙂

[Rickmer Krohn @RKrohn_]

Real-world online Reinforcement Learning from Vision + ForceTorque + Proprioception --> With MSDP we achieve ~90% success rate in only 35min of online training!

Checkout the Preprint here: https://t.co/bmme0ESX0v

Thanks to @FatAndFurious42 @GabrieleTiboni1 @GeorgiaChal!

[Reviewer3 @reviewer3com]

I’ve been frustrated for years by how slow, biased, and unpredictable peer review can be.

So I stopped complaining and built something different.

Reviewer 3 is your own panel of expert reviewers. They read your paper, highlight strengths and weaknesses, and now link every comment directly to your manuscript.

It started as a side project. Now it’s handled 9,000+ papers, and 88% of researchers say it’s as good or better than human review.

Here’s a sample review if you’d like to see what AI peer review looks like: https://t.co/q4rQPEm6v9

[Georgia Chalvatzaki @GeorgiaChal]

If you are at #ICLR2024 and are interested in Sim2Real for RL, reach out to @GabrieleTiboni1 to talk about our work DORAEMON that formulates the appropriate optimization problem over randomizing dynamics parameters to ensure policies become robust yet adaptive in the real world.

[Unitree @UnitreeRobotics]

Unitree H1 The World's First Full-size Motor Drive Humanoid Robot Flips on Ground.
Unitree H1 Deep Reinforcement Learning In-place Flipping !
#Unitree #UnitreeRobotics #AI #Robotics #Humanoidrobots #Worldrecord #Flips #EmbodiedAI #ArtificialIntelligence #Technology #Innovation

[AK @_akhaliq]

Learning Agile Soccer Skills for a Bipedal Robot with Deep Reinforcement Learning

investigated the application of Deep Reinforcement Learning (Deep RL) for low-cost, miniature humanoid hardware in a dynamic environment, showing the method can synthesize sophisticated and safe movement skills making up complex behavioral strategies in a simplified one-versus-one (1v1) soccer game

abs: https://t.co/AP5Lntzw1n
project page: https://t.co/I2AufxsW3b

[NVIDIA AI Developer @NVIDIAAIDev]

Excited to share our ICRA’23 @ieee_ras_icra work by @Adithya_Murali_

We scale up neural collision detection for object rearrangement with procedurally generated synthetic data.

Project: https://t.co/SGknqmtsmF
Video: https://t.co/LjYz3jaYRY
🧵👇

[Xiaolong Wang @xiaolonw]

🏗️ Policy Adaptation from Foundation Model Feedback #CVPR2023
https://t.co/l1vSFtLdYq

Instead of using foundation model as a pre-trained encoder (generator), we use it as a Teacher (discriminator) to tell where our policy did wrong and helps it adapts to new envs and tasks.

[Dhruv Batra @DhruvBatra_]

Last 4 years of @ai_habitat have been a steady march against moving goalposts:

— Model-free RL will never scale: Yes, it does with a fast sim, DD-PPO https://t.co/HStpkS9enq and VER https://t.co/N5psv0wIxv

— Performance in sim will never generalize to robots: Yes, it does and it is even good for model selection: https://t.co/qpAElJUFJP

— Modeling accurate physics is essential for sim2real: No, it isn’t for navigation: https://t.co/4mbyAs3GO0

— Navigation is a special case, sim2real won’t generalize to contact rich tasks: It does, see our work on mobile manipulation or the line of work on rapid motor adaptation.

[Pete @nonmayorpete]

Text to video.

It’s so over.

[Chelsea Finn @chelseabfinn]

We introduce a system for fine-grained robotic manipulation! 🤖

What’s new?
* We can control cheap robots to do surprisingly dexterous tasks
* New technique that allows robots to learn fine motor skills

A short thread 🧵