Collar off. Cup flips. Nothing spills. ✌️

Spoon placement is a two-hand job. One hand holds the Blizzard. The other picks up the spoon, places it, then pushes the ice cream off the collar and back into the cup. Small motion. Tight tolerance.

Three toppings, in the right order. Then straight into blending. This is where long-horizon execution matters: the robot has to keep track of what’s done, what’s next, and whether the current step actually worked.

Before filling the order, Sharpa North grabs another cup to catch the melted mix at the nozzle. Small step, but part of the real workflow.

Metal collar first, then one paper cup. The cup stack is a good tactile test: pull too hard and you take two, too light and nothing moves. Sharpa North adjusts grip as soon as the fingers make contact.

Sharpa North has been at work at DQ Shanghai. Making Blizzards for real customers, using the same setup and workflow as the rest of the store. We’ve learned a lot along the way - from customers, and from the DQ team working alongside Sharpa North. Here’s a closer look at one order, step by step. 👇

Sharpa at Silicon Valley 101 Alignment 2026 Summit. We brought Sharpa D01 and our latest dexterous hand, Sharpa W02, to the Bay Area — putting our latest work in robotic manipulation directly in front of the AI and robotics community. From full-body execution to dexterous hand manipulation, we’re showing how Sharpa is building the hardware foundation for robots to operate in the physical world. If you’re around the Bay Area, come see Sharpa in action. sharpa.com #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

Robotic dexterity, from system to hand. At #IROS2026, Sharpa brought together D01, W02 and AE01 to show different layers of real-world manipulation - fast and precise robotic motion, tactile dexterity, and intuitive teleoperation. Together, they showed how robotic capability translates into real physical interaction. Thanks to everyone who joined us! More from Sharpa: sharpa.com

Touch changes the interaction. At #IROS2026, even a light touch on Sharpa D01’s arms or chest can trigger a smooth response. D01 features full tactile coverage across interaction areas, enabling it to detect physical contact and respond accordingly. - 100 Hz tactile sampling rate - 0.1–20 N force sensing range - 0.2 N force resolution - 10 mm spatial resolution Tactile sensing gives robots another way to understand physical interaction — through contact itself. sharpa.com #IROS2026 #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

Catch it like you would with your own hand. With Sharpa W02, the operator reaches out and closes their hand as the toy comes in — and W02 mirrors the movement in real time to make the catch. A natural human gesture, translated directly into robotic motion. Compact, human-scale, and fully tactile, Sharpa W02 brings intuitive control to real-world manipulation. sharpa.com #IROS2026 #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

In-hand Rotation — Sharpa W01 SE With 30 FPS on-board inference and a dynamic tactile array, W01 SE coordinates all five fingers in real time to continuously rotate an object — smoothly and steadily. - 2 mm spatial resolution - 50 mN force sensing sensitivity - 30 FPS tactile sensing frame rate

Pen Spinning — Sharpa W01 With 180 FPS on-host inference, W01 coordinates all five fingers in real time to continuously spin and control the pen with speed and precision. Powered by Sharpa’s proprietary Dynamic Tactile Array: - >1,000 tactile pixels per fingertip - 1 mm spatial resolution - 20 mN pressure sensitivity

From uncapping to pouring, shaking, and mixing — 🍸Robot Bartending puts a full manipulation sequence into motion. Sharpa D01 is teleoperated through AE01, which maps the operator’s hand movements directly to the robot through a 22-DoF hand-glove twin design. On the execution side, D01 combines fine manipulation with high-speed motion and 1:1 human-scale spherical wrist joints for smooth, natural movement. - 22 DoF hand-glove twin design (AE01) - >10.5 m/s maximum end-effector speed (D01) - 1:1 human-scale spherical wrist joints (D01) 📍#IROS2026 | Booth 514 sharpa.com #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

How fast can a robot react? In the Catch the Stick Challenge, Sharpa D01 detects the falling stick, moves at high speed to intercept it, and uses precise control to secure the grasp. The challenge brings several capabilities together at once: - >10.5 m/s maximum end-effector speed - 0.2 mm repeatability - 1000 Hz communication frequency Fast motion alone isn’t enough. The robot has to reach the right place, at the right time, and complete the grasp with precision. See D01 live at #IROS2026. 📍 Booth 514 sharpa.com #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

A packed booth, live demonstrations, and plenty of conversations at #IROS2026. The Sharpa booth stayed busy throughout the day with: 🪄 Catch the Stick Challenge 🍸 Robot Bartending ✏️ Pen Spinning 🦾 Whole-Body Tactile Response 🧸 Catch the Toy 🔄 In-hand Rotation More live at Booth 514. sharpa.com #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

Robotic dexterity, in motion. Fast reactions. Precise manipulation. Rich physical interaction. A few moments from Day 1 at #IROS2026, where we’re putting Sharpa’s latest hardware through real-world manipulation tasks. Discover Sharpa’s three flagship products and learn more: sharpa.com 📍 Booth 514 #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

Sharpa will be co-hosting two competitions at IROS 2026. Powered by Sharpa's robot platforms and software, the 2nd ROCO and Origami challenges (cohosted by @BitRobotNetwork) brought together 48 teams from leading institutions around the world—including Carnegie Mellon University, UC Berkeley, KAIST and NUS. Across both competitions, teams will put robotic manipulation and dexterity to the test on Sharpa hardware through challenging real-world tasks. Follow the challenges for more: ROCO: @NtuLab38456 Origami: robotic-origami-challenge.gi…

Excited to announce: three new Sharpa products are making their debut at IROS — all developed in-house!🚀 Meet the lineup: D01 — general-purpose humanoid robot W02 — next-generation dexterous hand AE01 — exoskeleton data glove Discover what’s next in dexterous manipulation. See you in Pittsburgh! 📅 September 27–30 📍 Booth 514 · David L. Lawrence Convention Center Pittsburgh, PA, USA #IROS #Sharpa #Robotics #HumanoidRobotics #DexterousManipulation #DexterousHand #EmbodiedAI #RobotLearning #Teleoperation

We've been privileged to help some of the best teams advance general-purpose robotics models. Teams at @NVIDIARobotics working on #GR00T and EgoScale. Teams at @GoogleDeepMind shaping the new Gemini Robotics 2 model family. Researchers at @Stanford evolving RL policy training for dexterous manipulation. Researchers at @UCBerkeley pushing tactile policy learning forward. And many others we're proud to support quietly. That work has been a constant source of validation, on the need for dexterous hardware, and on the robustness and quality bar it has to meet. It's a sign the field is getting closer to real scaling laws, and closer to an inflection point in robotics autonomy. This is exactly why we've spent the past stretch strengthening our manufacturing and scaling production. More capacity. Faster lead times. More Wave units getting into the hands of the teams pushing the field forward. Reach out to know more (link in comments). #Robotics #EmbodiedAI #DexterousManipulation #RoboticsHand

Episode 4/4 — Same Kitchen. Same Equipment. Same Recipe. Real Customers. Building a general-purpose robot means it fits into a human environment, doing what a person does, not the other way around. At DQ: same kitchen, same equipment, same recipe. As @GoingBallistic5 put it, "absolutely no modifications whatsoever." No special dispenser, no custom handles, no rewritten process. That puts all the pressure on the robot. Retrofitting is the easy way out and the last thing a partner like DQ wants is to change what it's spent decades perfecting. North starts its first real shift at DQ Shanghai end of August. #Sharpa #SharpaNorth #EmbodiedAI #PhysicalAI #AIRobotics #DQChina #UsefulRobots @roydendsouza @mehrdad_Frimani

Did you know it takes #DQChina staff a full week of training to master blending a Blizzard? It's more complicated than it looks. Now imagine North having to learn the same move. A thin paper cup faces constantly shifting torque from thick, uneven ice cream. As @GoingBallistic5 put it, too little grip and it slips. Too much, and it crushes. It's hard for a parallel-jaw gripper to solve this: it only pinches from two sides. Wave doesn't pinch, but wraps. A five-finger grasp spreads the load and adjusts grip in real time as torque shifts, letting North run this exact motion autonomously. #SharpaWave #SharpaNorth #DexterousHands #EmbodiedAI #PhysicalAI #AIRobotics #DQChina #Blizzard #overthehorizon @roydendsouza @Hmorvaridi

And full video from #OverTheHorizon podcast here: piped.video/watch?v=QuEtIZdM…

Sharpa x DQ China full video teaser: piped.video/watch?v=143b_Pm7…

Episode 2 — Full Stack, Not Just a Hand Sharpa is often known for Wave, our dexterous hand. But a hand can't run a shift on its own. @GoingBallistic5 nailed it: North's DQ deployment shows "not just their hands, but the entire stack" — including our own embodied AI. Underneath: a world model predicting physics, CraftNet controlling motion, and within CraftNet our System 0, adjusting grip in real time through touch. No retrofits nor custom equipment. Our embodied AI adapts to the real world, not the other way around. Next: we put the stack to the test. #Sharpa #SharpaNorth #SharpaWave #EmbodiedAI #PhysicalAI #AIRobotics @roydendsouza @Hmorvaridi

Sharpa x DQ China full video teaser: piped.video/watch?v=143b_Pm7…

Full video from Over The Horizon: piped.video/watch?v=QuEtIZdM…

Episode 1 — We manufacture time by making robots useful This August, North steps behind the counter at a real DQ store in Shanghai; and it will be a full shift, with real customers, real orders, and of course, real ice cream. @GoingBallistic5 nailed it: North isn't a replacement for workers, it's a "substitute," filling in or augmenting the workforce where it's needed. The North robot adds capacity when teams are understaffed, late shifts are hard to fill, or repetitive work keeps people from more valuable tasks. @Hmorvaridi caught exactly what we meant with the football scene: people enjoy the match while North takes the shift. @roydendsouza linked it back to our mission: We manufacture time by making robots useful. Thanks, Royden, for pointing out that it is "the best worded mission" you've seen in a long time. Stay tune for the next episode! #Sharpa #SharpaNorth #UsefulRobots #EmbodiedAI #AIRobotics #overthehorizon

#SharpaWave helped train & test @GoogleDeepMind's Gemini Robotics 2. Backward compatible: the Wave adapts to any environment, object, or task. Same trash bags: no redesigned trash can. Same ziplock bags: no bulky boxes swapped in to make gripping easier. The hand fits the world, not the other way around. Fidelity matters too: Wave is designed to be anthropomorphic, with tactile resolution that captures the high quality data that human demonstration actually contains. #EmbodiedAI #PhysicalAI #Robotics #DexterousHands

The task of the #IROS2026OrigamiChallenge is to fold a traditional paper airplane in 6 folds. It's one of the hardest manipulation benchmarks around: paper is thin, deformable, easy to occlude, and unforgiving of contact timing. A successful policy has to coordinate two arms, two dexterous hands, tactile feedback, and multi-view vision over a long sequential horizon.

We're releasing the Origami Dataset we prepared for the #RoboticOrigamiChallenge we co-organize with @chris_j_paxton @micoolcho, @DJiafei and @lightwheel for #IROS2026! It's one of the largest tactile datasets out there and it's built on #SharpaWave, our 22-DoF five-finger tactile hand. It shows both head & wrist camera views; and records deformation and tactile patterns from the fingertips as the paper folding task unfolds. What's in the dataset: 1. 682 episodes across 51 collection seasons, 4.76M frames at 30 FPS 2. 6 synchronized camera views plus 3. 10 fingertips x 6-axis force/torque signals, synchronized frame by frame with vision and motion 4. 2 tactile video formats: a raw sensor stream for maximum fidelity, and a lighter deformation map 5. 65D joint-space state and action across two arms and two dexterous hands 6. Full LeRobot 3.0 coverage across all seasons, ready to drop into existing pipelines for long-horizon imitation learning, and/or visual-tactile representation learning Dataset on Hugging Face: huggingface.co/datasets/Shar… Wave URDF & tactile assets: sharpa.com/pages/open-source #RoboticOrigamiChallenge #HuggingFace #Robotics #TactileSensing #EmbodiedAI #IROS2026 #TactilePolicies #Robots #PhysicalAI #datasets

Most robot teleoperation copies hand motion but not "feel" so the robot crushes or drops whatever it's holding, and you throw the demo away. A new @NTUsg paper (from Research Fellow Haichao Liu, Dr. Ziwei Wang & team) addresses this: the robot reads its own fingertip tactile feedback and auto-corrects grip force in real time, while the human stays in control of the motion. Result: success on a hard assembly task went from 4% → 97%, using less than half the force. The bottleneck in dexterous robots training is also about collecting clean data where nothing got crushed along the way. This research help closing the gap. So do the #SharpaWave, that enable high quality data collection for complex manipulation tasks, like screwing. Link to the paper in comments! #DexterousManipulation #TactileSensing #AIRobotics #NTU #Dexteroushands #Robots #Teleoperation

Your first robot-handmade ice cream🤖🍦. Delivered to you by #SharpaNorth & DQ China this August in Shanghai. Autonomous. End-to-end. Full shift. Real store. Real customers. Real orders. #Sharpa #SharpaNorth #Shanghai #SharpaWave #AIRobotics #EmbodiedAI #PhysicalAI #robots #UsefulRobots #WorldModel #Blizzard #Oreocrumbs #icecream #worldcup

Sharpa at #RSS2026 in Sydney! Two North robots and our Wave hands sparked great conversations with the community working on manipulation and deployment bottlenecks. — Dr. Yunfang Yang on the Sharpa platform: hands, humanoid, AI models — Kaifeng Zhang, our Head of Academia, on our latest research — Met some of the best AI research talent in the field Building embodied AI? Come talk to us. #RSS2026 #AIRobotics #DexterousManipulation #SharpaWave #EmbodiedAI #SharpaNorth #VLA #PhysicalAI

🚀 We're coming to RSS 2026 in Sydney! Come to our booth near the main entrance to see how our dexterity-first technologies enable robust real-world productivity. We're looking forward to meeting researchers, engineers and founders from across the AI robotics community. 📍 Main Entrance 📅 July 13–16 📍 ICC – Sydney Stop by and say hello—we'd love to connect. #Sharpa #RSS2026 #Robotics #EmbodiedAI #DexterousManipulation #TactileSensing #HumanoidRobotics

NVIDIA's new CHORD framework teaches robot hands to manipulate objects by matching how human contact moves an object. Learned from human demos and transferred to real dexterous hands. Look at the high success rates: 82.12% success across 1,831 contact-rich tasks; 90.77% success on whole-body manipulation. With policies deployed from simulation. That's the #SharpaWave in the loop :) Project: nvidia-isaac.github.io/video… Paper: nvidia-isaac.github.io/video… #PhysicalAI #DexterousManipulation #RobotLearning #NVIDIA #EmbodiedAI #Robotics #Dexteroushands

Before a robot can perfect assembly, it needs to learn to play. The team behind SimToolReal @kushalk_ @tylerlum23 @leto__jean @KarenJLiu published another cool paper! Play2Perfect pretrains on diverse, task-agnostic play (grasp, reorient, reach, etc), then finetunes on sparse-reward assembly. Result: 33× sample efficiency vs. training from scratch, and zero-shot sim-to-real down to 0.5mm clearance. Peg insertion, screwing, multi-part assembly: all running at 60Hz, real speed, real hardware. And when a grasp slips, the policy doesn't stop, it recovers and keeps going. The Sharpa Wave responded present again ;) Project: play2perfect.github.io #Robotics #SharpaWave #Sharpa #EmbodiedAI #DexterousManipulation #RobotLearning

What if the next leap in robot manipulation comes from touch, not just vision? To get there, foundation models need to understand tactile feedback the way they understand images and language. And tactile policies cannot be locked to specific hardware (that makes real-world deployments & maintenance quite complicated). FTP-1 solves both. One of the 1st foundation model for touch. 21 sensors. ~3,000 hours of data. Transfers to hardware it has never seen before. +17% on known hardware. +31% on never-seen hardware. We're proud this research led by @michaelyuancb ran on #SharpaNorth, #SharpaWave hands, and our DTC sensors. Special thanks to the teams at @Tsinghua_Uni , @UCBerkeley , @ETH , and @sjtu1896. Project page: ftp1-policy.github.io/

North recently tried ironing. This week: shirt folding. 👕 Unlike ironing, folding is a skill even the most reluctant of our engineers have mastered. Notice how smooth the hand movements are. Clothes deserve some respect. 😉

The dataset was collected using two Sharpa Wave hands, using our high-resolution Dynamic Tactile Arrays on each fingertip to capture the dynamic touch signals needed for contact-rich manipulation. T-Rex proves that anthropomorphic hardware and tactile-reactive software are co-dependent :) Amazing work @Dantong_Niu ! P.S. The Wave's Isaac Sim URDF/USD assets and tactile parameters are available here if you wanna take a closer look: sharpa.com/pages/downloads

Robots need to feel the world to operate in it. Most manipulation policies today are tactile-blind. They either cannot interpret high-frequency tactile signals or treat them as a static channel. And the field lacks enough touch-rich datasets to train tactile-reactive policies at scale. T-Rex was built to answer both. @Dantong_Niu and team, advised by @DrJimFan, @drfeifei, @JitendraMalikCV, @pabbeel @trevordarrell have tested whether a robot policy can react to high-frequency tactile signals the way human hands do, without giving up the generalization power of modern VLAs. The result: 65% average success rate across 12 real-world tasks. +30 absolute points over the strongest baseline. In one year, tactile VLAs have gone from promising to outperforming non-tactile baselines like pi0.5 on dexterous tasks. #Robotics #SharpaWave #Sharpa #EmbodiedAI #DexterousManipulation #TactileSensing #RobotLearning Project link: tactile-rex.github.io/

We can use videos from the internet to teach robots! Do as I Do, from @bhawna_paliwal_ , @HarithejaE , and Willian Liang at UC Berkeley — advised by @pabbeel, @notmahi, and @JitendraMalikCV, used an algorithm that reconstructs hand-object interactions from monocular RGB video and retargets them into real, executable trajectories for multi-fingered dexterous hands. Just using "low quality" video footage of humans doing tasks. No sensors. The Sharpa Wave robot hand being anthropomorphic, it matches human kinematics. Not only that works, but at fast speed, too! Congrats to the team, that's super exciting! Project: do-as-i-do.com/ #Robotics #SharpaWave #Sharpa #EmbodiedAI #DexterousManipulation #RobotLearning

#ICRA2026 highlights: - @leto__jean and the @NVIDIARobotics team featured the Sharpa Wave in their SimToolReal and EgoScale talks 🙏 - We caught up with researchers and partners from around the world - North handed out Austrian chocolates 🍫 - There were many kids in the audience. Austria starts teaching robotics early 👀 See you at RSS! #ICRA2026 #AIRobotics #Robotics #IEEE #DexterousManipulation #TactileAI #Vienna #SharpaNorth #SharpaWave #DexterousHand

Question to the community: if a robot could iron your clothes, would you wear shirts? Asking for our engineers🙃.