Current Path：RobotHOT Briefs Dobot, CAS institutes' MoE paper accepted by EMNLP 2026

Dobot, CAS institutes' MoE paper accepted by EMNLP 2026

2026-09-04 Category: Briefs Comments(0)

A research paper co-authored by Dobot, the Institute of Computing Technology of the Chinese Academy of Sciences, the Nanjing College of the University of Chinese Academy of Sciences, the Nanjing Information High-Speed Rail Research Institute, and two other CAS-affiliated schools has been accepted for publication at the Findings of EMNLP 2026, the top-tier international conference on computational linguistics and natural language processing, which will take place in Budapest from 24–29 October. The study, titled “Meta-Learning Where to Allocate Experts: Task-Conditioned Layer-Wise Compression for MoEs,” proposes a novel compression method for mixture-of-experts models, marking a collaborative research achievement among the Chinese academic and industrial partners.

Like(0)
Share To

Related Articles

Comments Get first!

Cancel

Briefs

  • Autonomous first place: HKUST and GENISOM AI win with Gangbeng L2

    A team from the Hong Kong University of Science and Technology, working with GENISOM AI, won first place in the autonomous group for quadruped-wheeled robots at the 2026 Robot Warrior Challenge Autumn Competition in Beijing, the company said. GENISOM AI said its “GENISOM Gangbeng L2” robot, equipped with a robotic arm, competed in the autonomous operation category. The company said the result followed its “GENISOM Tongchui M1” platform’s six awards in the autonomous navigation category at the spring event, and showed its position in autonomous operation in complex, unstructured environments.

  • China's Guoyu Xingchen raises nearly 100 mln yuan in Series A funding

    Chinese space robotics startup Guoyu Xingchen has raised nearly 100 million yuan ($14 million) in a Series A round jointly invested by Yijing Capital, Fosun Chuangfu and Luyao Capital, the company said. The proceeds will focus on building space robot clusters and expanding commercialization, as the company seeks to accelerate China's in-orbit services from technical validation to scaled commercial delivery.

  • BYD granted design patent for embodied AI robot

    Chinese automaker BYD has been granted a design patent for an “embodied intelligent robot,” according to a patent abstract. The design covers a robot capable of walking, balancing and interacting, with potential applications in service, education and industrial settings.

  • China issues guidelines to speed tech transfer in consumer goods, including brain-computer interfaces and embodied AI

    China’s Commerce Ministry and eight other government departments have issued guidelines on accelerating technology-driven commerce, saying the country will cultivate technology transfer and commercialization institutions in the consumer goods sector. The guidelines, jointly released by the Commerce Ministry, Cyberspace Administration of China, National Development and Reform Commission, Ministry of Education, Ministry of Science and Technology, Ministry of Industry and Information Technology, Ministry of Culture and Tourism, People’s Bank of China and State Administration for Market Regulation, call for speeding the market conversion of frontier technologies including brain-computer interfaces and embodied intelligence.

  • China's DareRobot raises nine-figure yuan in Pre-A+ funding round

    Chinese industrial robotics startup DareRobot said it has raised a nine-figure yuan amount in a Pre-A+ funding round, without disclosing the exact figure. Investors included Fosun RuiZheng, Qingshi Asset Management, Zhengxi Capital, Xigao Investment, Bozhong Precision Industry and Suzhou Venture Capital, the company said. DareRobot, founded in 2022, makes intelligent cutting, grinding, welding and coating robots for heavy-industry sectors including defense, steel, marine and shipbuilding, and energy equipment. It targets automation in complex workpieces and high-mix, low-volume production. The proceeds will mainly fund research and development of core embodied-AI technologies for heavy industry, robot product upgrades and market expansion, the company said.

  • Poxiao AI, Tinno Mobile launch full-body multimodal data collection system

    Chinese embodied-AI company Poxiao AI and handset maker Tinno Mobile have jointly released SeePhys Hub, a full-body multimodal data collection system. The system comprises a four-camera head ring, wrist bands, waist and leg motion trackers and tactile gloves, the companies said. It can simultaneously collect first-person visual, two-handed tactile and full-body motion data, filling gaps in conventional ego-data collection, which struggles to capture body movement and contact pressure. A companion data-processing platform converts the collected content into multimodal training data for robot learning, they said.

  • China's Youibot clears CSRC filing for Hong Kong IPO, eyes first mobile-manipulation robot listing

    China's securities regulator on Oct. 16 published a filing notice for Hefei Youibot Robotics Co., Ltd.'s overseas listing and full circulation of domestic unlisted shares, clearing a precondition for a Hong Kong Stock Exchange hearing and putting the company on track to become the first listed mobile-manipulation robot maker. Youibot plans to issue no more than 49,477,800 overseas-listed ordinary shares and list on the Hong Kong Stock Exchange, according to the notice. Thirty-seven shareholders plan to convert a combined 243,804,260 domestic unlisted shares into overseas-listed shares, achieving "full circulation."

  • AGIBOT, Ant Afu deploy smart triage robot at Peking University People's Hospital

    The smart triage project jointly developed by AGIBOT and Ant Afu has been deployed at Peking University People's Hospital, with Lingxi X2 serving as a "silicon-based receptionist" in outpatient services. The robot can provide patients and visitors with department guidance, consultation process information, answers to medical insurance policy questions and on-site navigation. By connecting to the hospital's private knowledge base, Lingxi X2 can also deliver standardized medical guidance based on in-hospital information. Over 66 working days, it completed more than 46,000 interactions, averaging about 707 per day.
MORE

Hot Topics

RobotHOT - China & Global Robotics Insights

RobotHOT brings China & Global Robotics Insights. Explore robotics, embodied‑intelligence updates, startup news and industry analysis for professionals worldwide.

About USContact US