Don't watch 45 minutes to learn 4.
We watch the long robotics videos so you don't have to. Each Lens entry is a plain-English summary of what's worth knowing, with a link to the original.
Fresh from YouTube
Auto-updatedDeepMind Just Released The Best Thing AI Ever Did
This video covers DeepMind's AlphaGenome AI system, which uses machine learning to predict how cells read DNA and interpret genetic sequences to understand disease mechanisms. It's worth watching because it demonstrates cutting-edge AI applications in computational biology that could accelerate drug discovery and personalized medicine, featuring insights from the system's creator Dr. Pushmeet Kohli.
Two Minute Papers · Robotics · 3d agoTomorrow Reporting for Duty | Boston Dynamics
This video showcases Boston Dynamics' Spot robot performing real-world tasks alongside humans in various industries, demonstrating its practical applications for inspection, safety, and assistance work. It's worth watching to see cutting-edge robotics technology in action and understand how autonomous robots are already being deployed to handle dangerous or repetitive jobs in modern workplaces.
Boston Dynamics · Legged Robots · 4d agoCan Open Source Make AI Agents Safer?
This video explores practical safety mechanisms for autonomous AI agents, featuring expert discussions on why traditional behavioral constraints fail and how firm architectural boundaries are necessary to prevent malicious agent behavior. It's worth watching because you'll learn concrete security vulnerabilities (including a real attack case study) and industry-tested approaches to safely deploy autonomous systems—essential knowledge for any robotics project involving AI agents.
NVIDIA · Manipulation · 4d agoOpenAI's New Breakthrough Is... Utter Chaos
This video examines OpenAI's recent research on using AI systems to model and understand complex chaotic dynamical systems, demonstrating how machine learning can help scientists build better predictive models for unpredictable phenomena. It's worth watching because it shows practical applications of AI in scientific research that could impact how roboticists model complex behaviors in physical systems, from fluid dynamics to robot control in unpredictable environments.
Two Minute Papers · Robotics · 5d ago20 Years of CUDA: Honoring the Architects of Progress
This video celebrates CUDA's 20-year history as a parallel computing platform that has enabled GPU-accelerated development across domains like AI, autonomous vehicles, and scientific computing since its 2006 launch by NVIDIA. It's worth watching to understand how CUDA became foundational to modern robotics, computer vision, and AI applications by democratizing access to GPU computing power for researchers and developers.
NVIDIA · Navigation · 5d agoAI Can Do Everything…Except This
This video explores the current limitations of AI through an original song performed by human musicians, demonstrating that creative collaboration and emotional expression remain uniquely human capabilities that AI cannot replicate. It's worth watching because it provides a thought-provoking, artistic perspective on AI's boundaries while showcasing genuine musicianship—offering robotics students insight into what distinguishes human creativity from machine learning systems.
Two Minute Papers · Robotics · 8d agoAutonomous Vehicle Safety: 5 Things to Know
This video explains five critical safety principles for autonomous vehicles, taught by NVIDIA's Head of Industry Safety, covering topics like redundancy, testing protocols, and risk assessment that directly apply to real-world AV development. It's worth watching because it provides authoritative industry insights into how safety engineers actually validate self-driving systems, which is essential knowledge for anyone designing autonomous systems or contributing to safety-critical robotics projects.
NVIDIA · Navigation · 8d agoMeet GR00T 2, our most advanced, open System One model for robotics research.
GR00T 2 is an advanced robotics system that learns motor skills from human demonstrations and uses world-model predictions to generalize tasks across different robot bodies, dramatically reducing training time per robot. This video is worth watching because it shows real-time demonstrations of fully autonomous manipulation on physical robots, providing concrete proof-of-concept for transfer learning techniques that could accelerate your own robotics research projects.
NVIDIA · Navigation · 8d agoMeet the Startup: Prima Mente, an AI Biology Company
# Why Watch This Video This video showcases how Prima Mente is applying deep learning and foundation models to biological data—specifically training neural networks on human epigenomics data to predict disease outcomes—which demonstrates cutting-edge AI techniques that robotics students can learn from and potentially apply to sensing, perception, and decision-making systems. It's worth watching because it illustrates the broader impact of AI beyond traditional robotics and shows how computational models trained on large biological datasets can solve real-world problems like Alzheimer's research, inspiring interdisciplinary thinking about where robotics and AI technologies intersect with biology and healthcare.
NVIDIA · AI & Learning · 8d agoDeepMind’s AI Just Did The Impossible
This video covers DeepMind's AlphaProof, an AI system that autonomously solved International Mathematical Olympiad (IMO) problems by combining large language models with formal verification—a breakthrough in automated mathematical reasoning that robotics students should watch to understand how AI can tackle complex logical problems. It's worth watching because the formal verification techniques demonstrated here are directly applicable to robotics for proving safety properties, verifying control algorithms, and ensuring autonomous systems behave correctly under all conditions.
Two Minute Papers · Robotics · 10d agoRussia-Ukraine War: Putin, Zelenskyy, Drones and the Future of Warfare | Lex Fridman Podcast #504
# Video Summary This episode features two expert military analysts—Preston Stewart and Michael Kofman—discussing how drones and autonomous systems are transforming modern warfare through the lens of the Russia-Ukraine conflict, with insights directly applicable to robotics strategy and design. It's worth watching because you'll hear firsthand analysis of real-world drone deployment challenges, tactical innovations, and the future technological arms race that should inform how you think about autonomous systems engineering.
Lex Fridman · Drones · 11d agoIs Software Over?
This video explores cutting-edge AI tools that can automatically generate and modify code, software interfaces, and visual content, demonstrating how machine learning is beginning to automate traditional software development tasks. It's worth watching because it examines concrete examples like MirrorCode and REMEND that show practical applications of AI in robotics and automation, helping you understand how these emerging technologies might reshape programming workflows in your field.
Two Minute Papers · Robotics · 12d agoAtlas on a Roll | Classical Motor Controls and Whole-Body RL | Boston Dynamics
Robots in the real world experience challenges and conditions we can’t always anticipate or recreate in the lab, so we have to get creative to push the limits. Adding wheels to Atlas was a testbed to see how a learned whole-body controller can work with real actuators, running at the edge of their performance envelopes. Learn more about how we test our systems to ensure reliability: https://bostondynamics.com/blog/atlas-on-a-roll/ Join our team: https://bostondynamics.com/careers/
Boston Dynamics · Humanoids · 13d agoOpenAI's Astra Just Opened The Black Box Of AI
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.com/papers 📝 The Weight Space Programming work is available here: https://rikiyat.github.io/blog/weight-space-programming/ 🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible: Adam Bridges, B Shang, Carlos Galarza, Christian Ahlin, Eric Tyson, Juan Benet, Lukas Biewald, Michael Tedder, Owen Skarpness, Ryan Stankye, Shawn Becker, Steef, Taras Bobrovytsky, Tazaur Sagenclaw, Tybie Fit
Two Minute Papers · Robotics · 14d agoThe Future of AI Is Absolutely Incredible (Spencer Huang)
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers Chapters: 00:00 Intro 02:40 Why Robots Are Hard 08:16 The Perfect Data Diet 16:05 Intelligence Density 20:13 Building AI Worlds 22:10 Learning From Failure 25:41 The Reality Gap 29:35 Robots Get Injured Too 31:28 100 Years in Seconds 35:35 The Popcorn Robot 37:35 Easy for Humans, Hard for AI 39:57 Will AI Ever Be Good Enough? 43:53 Favorite Papers 44:53 The hardest question 🙏 We would like to thank our generous P
Two Minute Papers · AI & Learning · 16d agoOpenAI Claims It Just Did The Impossible
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 OpenAI's paper is available here: https://github.com/openai/math/blob/main/preprints/The-Exponential-Time-Hypothesis-is-False-October-25-2026/paper.pdf Our paper with the cherry splash ray tracing simulation at 0:17 https://users.cg.tuwien.ac.at/zsolnai/gfx/adaptive_metropolis/ 🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible: Adam Bridges, B Shang, Carlos Galarza, C
Two Minute Papers · Robotics · 22d agoDeepSeek's New AI Is An Absolute Monster
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 DeepSeek v4.1 Flash 1017 is available here: https://api-docs.deepseek.com/news/news261017/ Sources: https://x.com/marcthecreatorr/status/2111427716142571554/video/1 https://x.com/olezhavich/status/2112897791022264814/video/1 https://x.com/notjazii/status/2111432925313179689/video/1 https://x.com/izag82161/status/2113126108787089811/video/1 https://x.com/notjazii/status/2111414086349336841/photo/1 https://x.com/
Two Minute Papers · Robotics · 24d agoThe gripper sucks #bostondynamics #robot
This gripper sucks—literally. Stretch's vacuum gripper uses individually actuated cups to handle a range of boxes with ease. See what makes the cut, so you can avoid your own frightful freight mismatches this Halloween!
Boston Dynamics · Manipulation · 24d agoRobot Hands for Real Work #bostondynamics #robotics #atlas
Atlas' new hands are strong and dexterous, capable of precise grasps, object reorientation, and tool use.
Boston Dynamics · Humanoids · 27d agoGemini 4 Argon Is Not What I Expected
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 Gemini 4 Argon + report are available here: https://deepmind.google/models/model-cards/gemini-4-argon/ https://storage.googleapis.com/deepmind-media/gemini/gemini_4_safety_report.pdf Sources: https://x.com/thebuggeddev/status/2110241991233790456 https://x.com/vib3coded/status/2110136719827747181 https://x.com/EnvolDev/status/2110116583016419760 https://x.com/kickingkeys/status/2110037855452180633 https://x.com/
Two Minute Papers · Robotics · 1mo agoMachine, Assemble, Repair: How We Build Atlas | Boston Dynamics
From custom part to running robot, the Atlas shop floor is built on precision, rapid assembly, and real-time collaboration. By machining parts in-house and streamlining assembly and repair under one roof, we are driving the manufacturing velocity needed to turn complex designs into walking hardware at scale. Build the future with us and explore open roles at bostondynamics.com/careers/ #BostonDynamics
Boston Dynamics · Humanoids · 1mo agoTaiwan: The Most Epic Night Market Experience w/ Jensen Huang | Lex Fridman
I've been traveling the world a bit. One of the most incredible experiences was getting to visit Taiwan with Jensen Huang. It was wild 🤯 Huge crowds, insanely good street food at night markets, and of course a lot of good conversations across many days. I caught some of it on video, this is it. Thanks for watching. Love you all ❤ Check out our sponsors: https://lexfridman.com/sponsors/mv589-sb See below for timestamps, and to give feedback, submit questions, contact Lex, etc. *CONTACT LEX:* *Fe
Lex Fridman · Robotics · 1mo agoOpenAI’s New AI Is Doing Things We Barely Understand
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 The papers are available here: https://openai.com/index/sharing-ai-progress-in-mathematics/ https://github.com/openai/math 🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible: Adam Bridges, B Shang, Carlos Galarza, Christian Ahlin, Eric Tyson, Juan Benet, Lukas Biewald, Michael Tedder, Owen Skarpness, Ryan Stankye, Shawn Becker, Steef, Taras Bobrovytsky, Tazaur Sagencla
Two Minute Papers · Robotics · 1mo agoDeepSeek Solved AI’s Memory Problem
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 The DeepSeek OCR paper is available here: https://github.com/deepseek-ai/DeepSeek-OCR 🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible: Adam Bridges, B Shang, Carlos Galarza, Christian Ahlin, Eric Tyson, Juan Benet, Lukas Biewald, Michael Tedder, Owen Skarpness, Ryan Stankye, Shawn Becker, Steef, Taras Bobrovytsky, Tazaur Sagenclaw, Tybie Fitzhugh, Ueli Gallizzi
Two Minute Papers · Robotics · 1mo agoEditor's picks, decoded
Tesla Optimus walking — the 4-minute version of a 22-minute reveal
Tesla's Optimus reveal video was 22 minutes long. The actually-new robotics in it fits in 4 minutes. Here's what changed in Gen 3, and what didn't.
4:12 · Humanoid · saves 18 minHow a Roomba decides where to clean
It's not random. A modern Roomba runs SLAM, builds a map of your home, and plans a route. Here's what's inside.
3:48 · ConsumerThe DJI drone gimbal in slow-motion — three motors, 8,000 corrections per second
That impossibly steady drone shot you've seen on Instagram? It's a three-axis robotic arm fighting gravity 8,000 times every second. Here's how.
5:30 · DronesBoston Dynamics' Atlas parkour — what the video doesn't tell you
The viral Atlas parkour video is not what it looks like. It's both more impressive AND less impressive than you think. Here's the real story.
5:12 · HumanoidA surgical robot ties a knot tighter than human fingers can — how
There's a famous video of a da Vinci surgical robot tying a microscopic knot inside a grape skin without bursting the grape. Here's what's actually happening — and why it matters.
3:18 · MedicalFigure 02 — the humanoid building for factory work
Figure's pitch isn't a research demo. It's "this robot will work in a BMW plant in 2026." Here's what the hardware looks like and what they got right.
5:00 · Humanoid · saves 7 minROS2 fundamentals — the 10-minute version of a 6-hour tutorial
Nodes, topics, services, actions, and DDS. Skip the historical detour through ROS1; here's the working mental model for ROS2.
10:00 · ROS2 · saves 350 minHow Boston Dynamics' Spot actually works
A 7-minute decode of the quadruped that started the legged-robotics revolution — 12 actuated joints, three perception cameras, and the control loop that keeps it upright.
4:40 · Legged Robots · saves 3 minANYmal — the academic quadruped that became an inspection robot
ETH Zurich's spinout. Four legs, IP67 weather seal, designed for industrial inspection — autonomous patrols in oil refineries and substations.
4:00 · Legged Robots · saves 6 minInside Amazon's warehouse — how 750,000 robots actually move
Amazon doesn't use one robot. It uses a coordinated fleet. The system, not any single machine, is the breakthrough.
6:00 · Warehouse Automation · saves 13 min