Skip to content
#Opinions
Apr 28Tue
  1. KNOWIN 诺因智能 · 微信公众号43

    KNOWIN founder explains generative embodied intelligence technology path at POWER Robot main venue

    KNOWIN founder and CEO Li Yinchuan delivered a keynote at the main venue of the 2026 POWER Robot Future Conference, explaining a technology path that uses generative AI to rebuild the data paradigm for embodied intelligence and arguing for a shift from imitation to generation. The talk estimated that, benchmarked against GPT-3.5's 4T tokens training scale, embodied intelligence needs roughly 4 million hours of effective data; real-machine teleoperation would cost about 20-32 billion yuan, while a low-frame-rate video generation path could bring that down to 160-320 million yuan. KNOWIN has chosen the generative route, targeting deployment in home scenarios.

Apr 27Mon
  1. 线性资本65

    Wujie Dynamics completes Angel++ funding round, cumulative funding exceeds $200 million and signs over 500 million yuan in overseas orders

    General-purpose embodied intelligent robotics company Wujie Dynamics announced the completion of its Angel++ funding round, jointly invested by Envision Technology Group and the Beijing Artificial Intelligence Industry Investment Fund, with existing shareholders including Sequoia China, Linear Capital, and Hillhouse Venture Capital participating. Cumulative Angel-round funding exceeds $200 million, and the Angel+++ round is nearing completion.

    Why it matters: The original text provides three verifiable pieces of information at once—funding, orders, and personnel—allowing readers to judge the company's commercialization pace and technology-route choices.

Apr 26Sun
  1. Chelsea Finn23

    Finn's Three ICLR Workshop Talk Preview

    Tomorrow (Sunday) I will give three workshop talks at ICLR: - Self-improvement without reward bottlenecks, including meta-harness (10:30 AM, RSI workshop) - Emergent physical generalization, including π0.7 (11:40 AM, multimodal workshop) - Reinforcement learning for robustness (1:45 PM, CAO workshop)

Apr 24Fri
Apr 21Tue
Apr 18Sat
  1. IntBot · X14

    Co-founder Discusses the Frontier of Embodied AI

    Papers don't deploy. Demos don't scale. The gap between what works in a lab and what works in an airport terminal at 6 AM is the entire value of a company. Our co-founder David Yuan wrote in Forbes about why AI's next frontier is not in the digital world, but in embodiment.

Apr 17Fri
  1. 线性资本50

    In Conversation with Manycore's Huang Xiaohuang: A 15-Year Startup Retrospective from Kujiale to Spatial Intelligence

    Linear Capital founder Wang Huai and Manycore Technology co-founder and chairman Huang Xiaohuang recorded the first episode of the video podcast "Between the Lines," reviewing 15 years of entrepreneurship and publishing a text excerpt on the day Manycore listed on the Hong Kong Stock Exchange (00068.HK). Huang Xiaohuang discussed how Manycore is entering spatial intelligence based on its accumulated indoor 3D data, positioning it as a bridge between large language models and the physical world, and introduced SpatialLM, open-sourced last March, which converts point clouds and 3D Gaussians into scripts that large language models can understand.

Apr 16Thu
  1. OFweek Robotics50

    Youlu Robotics' AI130 Patrol Sweeping Robot Tops China's Outdoor Sweeping Robot Shipments in One Year

    Youlu Robotics, with its AI130 patrol sweeping robot, rose to the top of China's 2025 outdoor sweeping robot shipments within one year. Its core product is the Master2000 embodied brain with 254 TOPS of computing power, which uses a "one brain, multiple forms" model to adapt to different robot forms such as patrol sweeping and security patrols, and it has developed its own end-to-end general-purpose LPLM multimodal model. The company's "three gates" strategy progressively moves from public residential-community scenarios into unit doors and homes, feeding real-world scenario data back into the brain to achieve "deployment-free, out-of-the-box" delivery.

Apr 9Thu
  1. New Scientist Robots19

    Ode to the Half-Broken and Luminous: Two New Sci-Fi Novels Explore Robots in Very Different Ways

    Ode to the Half-Broken and Silvia Park's debut Luminous are two new robot-themed science fiction novels read this month, with very different styles. In Suzanne Palmer's book, an old robot whose legs were stolen teams up with a half-dog half-robot, a human mechanic, and a group of liberated drones for an adventure; the tone is light and humorous, and it may suit younger readers. Silvia Park's story is set in a future 20 years after Korean reunification, where robots are everywhere, often abused, and sometimes dangerous, and the tone is heavy.

Apr 6Mon
  1. 云启资本55

    Yunqi x Tsinghua: Five Embodied AI Practitioners Discuss Physical AI Deployment in Roundtable

    At the Physical AI Frontier Exploration and Industrial Innovation Symposium hosted by Tsinghua University's Department of Electronic Engineering and others on March 28, Wang Yu moderated a roundtable forum where five alumni—Chen Yilun, Yao Maoqing, Wang Qian, Ding Wenbo, and Li Ziyi—discussed the technical bottlenecks and deployment of embodied intelligence.

Apr 2Thu
  1. Prime Movers Lab · X26

    Humanoid robot hardware costs fell 40% in one year

    Humanoid robot hardware costs fell 40% in just one year. In the physical AI race, the moat is not who can build the best robot, but who puts it into use first. Real operating records and actuarial data will determine the winners. 🤖🏭 Read our latest white paper: https://bit.ly/4s9VJKr

Apr 1Wed
  1. 普罗宇宙20

    Embodied Insight | Pro Universe President Ge Jin Invited to 2026 Shanghai-Jiaxing AI Industry Matchmaking Exchange Conference

    Pro Universe Robotics President Ge Jin was invited to attend the 2026 Shanghai-Jiaxing AI Industry Matchmaking Exchange Conference and spoke during the roundtable session, stating that the large-model dividend period is far from peaking and is shifting from "general-capability showmanship" to "deep cultivation of vertical scenarios." Pro Universe builds an embodied intelligence industrial large model on the foundation of real application data from industrial scenarios, insists on local deployment to ensure data security, and uses the "general-purpose robot body + quickly replaceable end effector" architecture of its "Dabai" robot to achieve line changeovers in seconds to minutes. During the event, Pro Universe Robotics successfully signed an agreement at the project signing ceremony in Pinghu Economic Development Zone.

Mar 26Thu
  1. 赛那德27

    iLoabot-X Wins Innovation Case Award at 2026 Logistics Technology Conference; SENAD Decodes the Path to Embodied AI Deployment

    SENAD's autonomous robot iLoabot-X won the Best Innovation Case Award at the 2026 Logistics Technology Conference. Company deputy general manager Han Hetian delivered a keynote speech at the loading and unloading technology forum of the conference in Huzhou, sharing the deployment logic of embodied AI in the logistics sector. Drawing on real operational data accumulated from vertical logistics scenarios, SENAD has built the logistics industry's first multimodal VLA vertical large model. The company proposes a path of "deep cultivation of vertical scenarios + data closed-loop iteration" and will continue to focus on the pain points of the final 20 meters of loading and unloading in logistics while iterating this VLA vertical large model.

Mar 20Fri
Mar 19Thu
  1. Wandercraft50

    Wandercraft warns of industrial data risks as foreign robots enter European factories, advocates European autonomous industry path via French-Japanese alliance

    Wandercraft published an article stating that when non-European-made humanoid robots are deployed in European factories and hospitals, they collect industrial and medical data that flows out via servers not controlled by Europe, while integrators cannot access the source code. The article cites data including China accounting for over 80% of the global humanoid robot market, Unitree announcing 5,500 units, and AgiBot having delivered 5,168 units, and draws analogies to the industry precedents of electric vehicles, photovoltaics, and batteries, arguing that Europe should pursue a path built on three pillars: software sovereignty, reciprocal industrial alliances, and a local market in strategic sectors.

Mar 18Wed
Mar 13Fri
  1. Hexagon · X15

    Hexagon CTO on human-machine collaboration in autonomous systems

    Advances in autonomous systems are shifting the focus from standalone machine capability to human-machine collaboration. Our CTO Burkhard Boeckem explains how this boosts industrial efficiency—and why human oversight remains essential. Read the full article: @TechRadarPro: https://hxgn.biz/416VFzS

Mar 11Wed
Mar 9Mon
Feb 25Wed
  1. 深创投28

    Shenzhen Capital Group Chairman Zuo Ding: Target Key Sectors, Do Not Fear Innovation Risks, Invest Real Money

    Speaking at Guangdong Province's high-quality development conference on February 24, Shenzhen Capital Group Chairman Zuo Ding said the firm will target key sectors in advanced manufacturing and modern services, investing real money to cultivate Guangdong's leading technology enterprises. As of the end of January 2026, Shenzhen Capital Group had managed total funds of more than 500 billion yuan across all categories and had directly invested in more than 1,800 companies, over 90% of them hard-tech firms. In December 2025, Shenzhen Capital Group helped bring a Guangdong-Hong Kong-Macao Greater Bay Area fund with total scale of 50.45 billion yuan to Shenzhen; it is the first national-level fund of funds undertaken by Guangdong Province.

Feb 23Mon
  1. Dexterity37

    Dexterity explains why Physical AI is hard

    Physical AI—AI that drives robots to complete physical tasks in the real world—is hard because the physical world is endlessly variable: boxes deform and environments change from shift to shift. A truck-loading robot must simultaneously solve dozens of AI problems, including perception, planning, collision-free dual-arm motion, and force control, while achieving 99.9995% safety confidence for each action.

Feb 10Tue
  1. 云启资本48

    Yunqi Research Report | Embodied Intelligence in Industrial Scenarios: An In-Depth Analysis of Capabilities and Applications

    Yunqi Capital released a research report analyzing the capability gains and deployment boundaries of embodied intelligence in manufacturing and logistics scenarios relative to traditional industrial robots. Traditional industrial robots can reach 80–150 cycles per minute, while current VLA end-to-end execution remains at the 10–15 cycles per minute level, making it difficult to meet sub-millimeter precision requirements. Among deployment directions, inspection is one of the most practically feasible, material handling has the greatest potential for scale, and precision assembly has the highest barrier; the viable form is Agent architecture scheduling plus vertical-domain operation model execution.

Feb 5Thu
  1. Wandercraft43

    Wandercraft CEO's view: How human-like should a humanoid robot be?

    Wandercraft co-founder and CEO Matthieu Masselin argues in an article that humanoid robots should aim for "human-machine compatibility" rather than "looking like humans." The company's robots kept a headless design when deployed at the Renault factory, with cameras initially located on the lower abdomen and now moved between the shoulders; the company is developing end effectors and manipulation capabilities for its next-generation product. He believes de-anthropomorphizing can reduce psychological friction and speed up deployment.

Feb 2Mon
  1. 线性资本62

    Interview with Tashi Intelligence's Chen Yilun: No VLA, no simulation—an embodied AI company's contrarian bets

    Linear Capital reposted LatePost's interview with Tashi Intelligence founder Chen Yilun, who laid out a technical path that differs from the mainstream: he rejects the VLA architecture that grows an action head on top of a VLM, and instead develops its own AWE (AI World Engine), aiming to represent physical quantities such as time, space, and force; it has also abandoned teleoperation data collection in favor of a self-developed collection device combining a wearable glove and a first-person-view camera.

    Why it matters: The interview systematically presents technical judgments and a data approach that differ from the mainstream VLA path, giving readers a basis for understanding the specific grounds of the divergence in embodied AI approaches.

Jan 16Fri
  1. Addverb Technologies11

    13 Major Advantages of Warehouse Automation in 2026

    The article lists 13 benefits of warehouse automation, covering the use of technologies such as autonomous mobile robots (AMRs), automated guided vehicles (AGVs), automated storage and retrieval systems, pick-to-light picking, RFID tags, and computer vision in material handling, sorting, picking, and inventory management.

Jan 13Tue
  1. 线性资本49

    After CES, what is the answer for AI moving into the physical world? — Linear Capital's AI consumer hardware discussion

    Linear Capital partner Yang Jun spoke with Ling Universe's Gu Jiawei, Ludens AI's Xue Lijun, and Bangtang Technology's Lou Xiaodu about AI consumer hardware, focusing on education, companionship, and health. Ling Universe's Xiaofang device shipped more than 20,000 units early on, validating the payment logic of parents buying an "entry ticket" to the AI era for their children; Bangtang Technology cut the customer acquisition cost of its "AI pastor" software from $1.5–2 to $0.3–0.5 and raised its renewal rate from 70% to 80%–90%, and on that basis decided to make a smart ring to enter mental health and religious scenarios.

Jan 12Mon
  1. Skild AI48

    Skild AI: Enabling robots to learn new skills by watching human videos

    Skild AI proposes overcoming the robot data bottleneck by watching human videos, arguing that teleoperation is limited by environment diversity and real-time collection scale and cannot reach the data volume required by foundation models. Its approach is cross-embodiment learning: with less than 1 hour of robot data for fine-tuning, the model can learn new skills from video demonstrations alone, mapping human "grasping" onto the execution actions of different robot embodiments.

Jan 8Thu
  1. 线性资本40

    In conversation with Horizon Robotics' Yu Kai: Buffett and Munger are absolute idols, and entrepreneurship is essentially a form of investment | Linear Talk

    Wang Huai, founder of Linear Capital, held a fireside chat with Yu Kai, founder and CEO of Horizon Robotics, reviewing Horizon's ten-year entrepreneurial journey and its life-or-death moments. Yu Kai said Buffett, Charlie Munger, and Duan Yongping are his absolute idols, and that entrepreneurship and life decisions are essentially both investments. Horizon follows a risk-averse strategy of "competing where there is no competition" and "not dancing at the edge of a cliff," and at the end of 2019, when it still had more than 3 billion in cash on its books, it cut its headcount by half.

Dec 31Wed
  1. 线性资本38

    Linear Capital's Wang Huai: The future of AI has only two doors, one in the US and one in China | Linear View

    In a keynote article for the AGM conference, Linear Capital founder and CEO Wang Huai argues that before AI's superpower there are only two doors: the US's 0-to-1 innovation and China's 1-to-100 scaling capability. Linear requires its team to subscribe to and use the best AI models from both the US and China (ChatGPT, Gemini, Claude, DeepSeek, Kimi, and others), with the company fully reimbursing the cost.

Dec 25Thu
  1. 线性资本30

    Original Aspiration, Patience, and Confidence: Lessons from 10 Years of Smart Glasses Ups and Downs | Linear Voice

    In a "Linear Tech π" sharing session, Zhao Weiqi, head of Rokid's global open ecosystem, broke down the underlying thinking behind Rokid's journey through a ten-year cycle from three dimensions: original aspiration, patience, and confidence. Over ten years, Rokid has accumulated more than 1,000 intellectual property items and 547 domestic and international patents, concentrated in human-computer interaction, optics, acoustics, spatial computing, and OS capabilities; the company uses a "211" work model, with two rounds of iteration per month, once every two weeks, one week for validation and one week for review.

Dec 23Tue
  1. RealMan/睿尔曼27

    Realman Founder Zheng Suibing Proposes Building a Remote Workforce Network at Corporate Social Responsibility Forum to Create a Sustainable "Real-Machine Data Engine"

    Zheng Suibing, founder and CEO of Realman, said on December 19 at the 2025 Corporate Social Responsibility Forum that the company is using a "remote workforce network" to let operators direct robots in real time to perform real tasks such as folding towels, utility tunnel inspections, and power control, while accumulating scenario data to feed embodied intelligence. Realman's integrated joint modules have annual production capacity exceeding 100,000 units, its robotic arms have passed CR L3 certification, and their mean time between failures has reached 50,000 hours. The company also co-founded the GPR open-source alliance in February 2023 and released a mobile composite robot development platform in 2024.

  2. Sunday Robotics · X18

    Memo acquires grasping intuition through datasets

    Traditional robots are like player pianos—they can only play the notes they are given. Memo is more like a top chef who has watched hundreds of chefs in various kitchens. Our diverse dataset gives Memo the "intuition" to grasp entirely new objects. This is a small step toward robots that can truly help in any home 🤖👨‍🍳

Dec 4Thu
Nov 19Wed
Nov 13Thu
Nov 7Fri
Nov 4Tue
  1. Holiday Robotics50

    Holiday Robotics' industrial humanoid robot Friday uses self-made tactile sensors costing about 30,000 won, with mass production next year

    Holiday Robotics has unveiled Friday, an industrial humanoid robot whose hands use a 500-gram, 20-DOF hydraulic ultralight design, with about 80% of their surface covered by self-made tactile sensors that can sense a force of 0.01N (about 1g), as well as direction and torque.

Oct 27Mon
  1. SIGRobotics · X7

    Team Gains a Lot from Embodied AI Hackathon

    Hello everyone! We had a wonderful time at the #EmbodiedAIHackathon embodied AI hackathon with @seeedstudio, @NVIDIARobotics, and @huggingface! We ran into many challenges and problems, but we feel we learned a lot and gained more experience in embodied AI.

Oct 24Fri