Pratik Kunapuli retweetledi
With these new robots from @EkaRobotics we might be approaching a ChatGPT moment for the physical world
cc: @PKunapuli
wired.com/story/when-rob…
English
Pratik Kunapuli
33 posts
Pratik Kunapuli
@PKunapuli
Research Engineer @EkaRobotics. CIS PhD UPenn. RL, Robotics, Dynamics and Control. NSF GRFP Fellow.
Philadelphia, PA Katılım Ağustos 2018
276 Takip Edilen179 Takipçiler
Pratik Kunapuli retweetledi
A lot of companies are currently chasing a "ChatGPT moment for robotics". Eka, founded by @pulkitology and @haarnoja, is taking a different approach to most, and it has already produced an eerie new kind of robot intelligence. (My latest piece for @WIRED)
wired.com/story/when-rob…
English
Excited to be able to share what I've been working on for the past 6 months!
@EkaRobotics is building something really special, and our robots don't have to sacrifice speed or dexterity for generalization
Pulkit Agrawal@pulkitology
Eka means unity -- “one,” in Sanskrit and “first” in Finnish. We’re building intelligence for the physical world in its native language: forces. Until now, robotics faced a tradeoff — generality or speed. The real world requires both. Robotics also faced a data problem. Our Vision–Force–Action (VFA) model — the first of its kind — breaks the generality-speed tradeoff and the data barrier. It's a new foundation uniting performance, generality, and safety for putting capable robots in everyone's hands. Today, I am excited to share our journey of pushing robots beyond human limits. Today, dexterity becomes scalable. Today, I welcome you to the Era of Eka. Co-founded with @haarnoja, and so thrilled and grateful to be working with a dream team at @EkaRobotics. Learn more: ekarobotics.com
English
Pratik Kunapuli retweetledi
Until now, robotics stopped where the human hand begins.
Strong, but not delicate.
Precise, but not adaptive.
Repetitive, but not creative.
The human hand wasn’t a benchmark — it was a boundary.
We are crossing it.
@EkaRobotics.
Coming soon.
English
Pratik Kunapuli retweetledi
💡Can robots autonomously design their own tools and figure out how to use them?
We present VLMgineer 🛠️, a framework that leverages Vision Language Models with Evolutionary Search to automatically generate and refine physical tool designs alongside corresponding robot action plans.
✨ VLMgineer can fully automate tool and action design with AI-driven physical creativity. No human intervention. No pre-defined templates or few-shot examples.
✨ VLMgineer outperforms human-specified designs and existing everyday tools.
✨ We let the VLM fully decide how to evolve designs.
Deep dive with me: 🧵
English
REGENT will be presented as an Oral at ICLR 2025, given to the top 1.8% of 11672 submissions!
Kaustubh Sridhar@_k_sridhar
Is scaling current agent architectures the most effective way to build generalist agents that can rapidly adapt? Introducing 👑REGENT👑, a generalist agent that can generalize to unseen robotics tasks and games via retrieval-augmentation and in-context learning.🧵
English
Thrilled to have won the best paper award!
Great work done by my co-authors: Jake Welde & Nishanth Rao, and advisors: @dineshjayaraman & @vijay_r_kumar
Check out the website w/ code: pratikkunapuli.github.io/EQTrackingCont…
Symmetry and Geometry in Neural Representations@neur_reps
The best paper award for the Neuroscience and Interpretability track goes to "Leveraging Symmetry to Accelerate Learning of Trajectory Tracking Controllers for Free-Flying Robotic Systems", presented by Pratik Kunapuli and Nishanth Arun Rao.
English
A massive thank you to my co-authors: Jake Welde & Nishanth Rao, as well as my advisors: @dineshjayaraman , and @vijay_r_kumar . Also, we deeply appreciate support from ARL DCIST, NSF, NVIDIA, and the NSF GRFP. 🙏
Let’s connect and discuss! See you at #NeurIPS2024!
English
📊 Key Findings:
- Exploiting symmetry significantly boosts sample efficiency.
- Reductions in state space lead to policies that generalize well to new trajectories.
- Even for underactuated systems where gravity "breaks symmetry" (quadrotors), we still get great tracking!
English
Thrilled to announce that I'll be presenting our paper, "Leveraging Symmetry to Accelerate Learning of Trajectory Tracking Controllers for Free-Flying Robotic Systems", at the NeurReps Workshop @NeurIPSConf! 🎉
Learn more: pratikkunapuli.github.io/EQTrackingCont…
A quick 🧵👇
English
Pratik Kunapuli retweetledi
Dear all,
Slack workspace icons are there to help to identify your workspace. Make it distinctive.
Thanks
English