Twitxy Saved

Shivam Vats

Verified account This account is verified. Learn more

@ShivaamVats

Postdoc @BrownBigAI
Previously: PhD @CMU_Robotics, Maths @IITKgp, Core developer @SymPy
Providence, RI shivamvats.com Joined February 2010
550 Following 849 Followers 267 Posts

Latest posts 5

Pinned post

Shivam Vats Verified account @ShivaamVats
Robots shouldn't stop learning once they're deployed. How should they self-improve in the real world?

📢 Our CoRL 2026 workshop, Learn@Deploy, is happening tomorrow morning!

📆 When: Thursday, November 12, 8:30 AM–12:30 PM
🌐 Website and schedule: learn-at-deploy.github.io
Shivam Vats Verified account @ShivaamVats Sep 8
How can robots keep learning once deployed in the messy, ever-changing real world?

Excited to announce our #CoRL2026 workshop on “Robot Learning at Deployment”!

We’ll explore how robots can continually adapt and improve under real-world constraints. 🧵
View quoted post on X
4 replies 2 reposts 22 likes 1.1K views View on X
Shivam Vats Verified account @ShivaamVats
How can robots keep learning once deployed in the messy, ever-changing real world?

Excited to announce our #CoRL2026 workshop on “Robot Learning at Deployment”!

We’ll explore how robots can continually adapt and improve under real-world constraints. 🧵
1 replies 9 reposts 36 likes 2.3K views View on X
Shivam Vats Verified account @ShivaamVats
Thanks for the shoutout @tomssilver! Really glad you liked the paper, especially since “Practice Makes Perfect” was one of the influences behind it :)

Lots more to do!
Tom Silver @tomssilver Aug 30
This week's #PaperILike is "Deliberate Practice: Learning Robot Skills under a Budget" (Vats et al., 2026).

Practice your robot skills in a provably optimal way. Big fan of this whole line; see also arxiv.org/abs/2209.13605 & arxiv.org/abs/2505.00490

PDF: arxiv.org/abs/2608.13415
View quoted post on X
0 replies 1 reposts 20 likes 1.3K views View on X
Shivam Vats Verified account @ShivaamVats
Super excited about this collaboration with @rai_inst that enables robots to continually learn from real-world experience without supervision or simulation.
Lakshita Dodeja Verified account @lakshitadodeja May 11
Can BC policies be quickly improved through real world experience?

Our new #RSS2026 paper proposes Q2RL, a method that bridges BC and RL for on-robot learning.

Q2RL improves BC policies by up to 3.75x with just 1-2 hours of online interaction!

So when life gives you BC, make Q-functions! 🍋

Details in thread 🧵
View quoted post on X
0 replies 3 reposts 9 likes 1.5K views View on X
Shivam Vats Verified account @ShivaamVats
Happy to share our new RAL paper on efficient exploration in reinforcement learning for robot manipulation.
Lakshita Dodeja Verified account @lakshitadodeja Mar 18
Residual RL is a powerful strategy for adapting a pretrained base policy, but it struggles when:
❌ Exploration is uncontrolled
❌ Base policies are stochastic
We tackle this in our new RAL paper “Accelerating Residual Reinforcement Learning with Uncertainty Estimation”
(1/5)
View quoted post on X