Log inSign up
user avatar
Pulkit Agrawal
@pulkitology
Co-Founder @EkaRobotics, Faculty @MIT
Cambridge, MA
people.csail.mit.edu/pulkitag/
Joined June 2009
260
Following
16.5K
Followers
  • Pinned
    user avatar
    Pulkit Agrawal
    @pulkitology
    Apr 29
    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
    Image
    00:00
    327K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Nov 17, 2023
    We have been developing a policy optimization method to push the limits of agility while conserving energy. Here is a sneak peak from our robots attempt to spin:
    Image
    00:00
    94K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    May 6, 2022
    This robot spins, this robot runs! Be it snow, be it gravel, be it bricks or a broken motor. We achieved a record speed of 3.9m/s on the Mini-Cheetah while maintaining robustness on varied terrains using reinforcement learning. Find out more: agility.csail.mit.edu #Robotics
    Image
    00:00
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Jan 30, 2022
    The MIT Mini Cheetah achieving a record speed of 4.1 m/s. It runs fast outside the lab too :) A teaser from ongoing work led by @gabe_mrgl and @EpisodeYang pushing the limits of agile locomotion.
    Image
    00:00
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Mar 4, 2025
    Overturning the next-token prediction is required for achieving general reasoning! We predict that RPT (Reward Pre-Training) will overtake GPT in the future -- similar to how AlphaZero overtook AlphaGo. Learn more: arxiv.org/pdf/2502.19402 🚨Our whitepaper, “General Reasoning
    Image
    53K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Feb 29, 2024
    Presenting a method for training models from SCRATCH using LoRA: 💡20x reduction in communication 💡3x savings in memory - Find out more: minyoungg.github.io/LTE/ - Code available to try out - Scaling to larger models ongoing - arxiv.org/pdf/2402.16828… led by Jacob Huh!
    Image
    72K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Oct 8, 2022
    What does our robot do on the weekend? Check out snippets from @gabe_mrgl and @JiYandong ... with reinforcement learning techniques leading to robust locomotion in the last couple of years, there is so much that can be done .. here is just a glimpse.
    Image
    00:00
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Feb 15, 2025
    You aren't doing robotics if you are not breaking some robots! uan.csail.mit.edu
    Image
    00:00
    27K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Apr 5, 2023
    Introducing DribbleBot: A robot that can dribble a soccer ball on diverse natural terrains. Be it snow, be it grass, be it pavement, or be it sand, it keeps going! 🤖 if the robot falls, it automatically recovers and keeps dribbling. 🤖⛹️‍♂️ gmargo11.github.io/dribblebot/
    Image
    00:00
    55K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Nov 18, 2022
    Presenting Visual Dexterity: Object re-orientation in its full generality! Single camera. Novel objects. Any orientation. Downward-facing hand that fights gravity. Real-time dynamic control. Open source setup. Learn more: taochenshh.github.io/projects/visua… Led by @taochenshh #robots #rl
    Image
    00:00
  • user avatar
    Pulkit Agrawal
    @pulkitology
    May 2, 2023
    Diffusion Models are powerful for decision making: 💡 Provide an alternative to RL by stitching trajectories without dynamic programming! 💡Compose skills without hand-design of heirarchies or task-planners!! Presenting Decision Diffusers at #ICLR2023: anuragajay.github.io/decision-diffu…
    Image
    GIF
    32K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Nov 11, 2021
    Congratulations to @taochenshh and Jie for winning the Best Paper Award at CoRL 2021 @corl_conf. It took a lot of grit, but was a lot of fun :)
    Image
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Nov 1, 2023
    Announcing open-source code for training locomotion controllers: github.com/Improbable-AI/… These robots run on ice, gravel, over large stones, stairs, under obstacles, on slippery surfaces, dance, recover from falls & more! Excited to see what your robots do! Led by @gabe_mrgl
    Image
    00:00
    25K
  • user avatar
    Pulkit Agrawal
    @pulkitology
    Nov 6, 2024
    Introducing DART: breaking the barriers for robotic data collection by enabling anyone, anywhere in the world to control robots without even having a robot. Just log into dexhub.ai to contribute and control robots at much lower fatigue and higher speed than
    Image
    00:00
    24K