Dyna Robotics Launches DYNA-2 Model Trained on 1 Million Hours of Human Video
- Published
- Source
- InterestingEngineering.com
Dyna Robotics has introduced its DYNA-2 World-Action Model, trained on over 1 million hours of human video. This innovative approach aims to address the challenges of teaching robots to perform physical tasks by leveraging human egocentric video instead of traditional robot action data.
The significance of this development lies in its potential to enhance task success rates in high-precision manufacturing, with Dyna reporting an increase from 20% to 80%-90% due to the extensive pre-training. The model's architecture allows for knowledge transfer across various robotic platforms, demonstrating adaptability with minimal fine-tuning.
Looking ahead, Dyna Robotics envisions DYNA-2 as a pivotal advancement in robotics, enabling robots to learn new tasks without extensive robot-specific training data. No further timeline was disclosed at the time of publication.
Original report
Humanoid robots trained on 1M hours of human video achieve up to 90% task success
InterestingEngineering.com · Neetika WalterThis briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
Related Suppliers
Dyna Robotics
US startup building embodied AI foundation models for dual-arm commercial robots, achieving 99%+ success in 24/7 deployments in restaurants and laundromats.
SolverX
South Korean AI-aided engineering startup that accelerates CAE simulation workflows using physics-informed AI models across automotive, aerospace, and.
Robótica Industrial Proyectos México
Mexican robot training and system integrator. 20+ years delivering ABB, FANUC, KUKA, Yaskawa courses and turnkey robotic cells for automotive industry.
Synthium
Synthium is a Singapore-based, Antler-backed startup that crowdsources gamified simulations to generate training data for humanoid robot AI models.
MediaMeta
Singapore's MediaMeta runs a nine-product AI platform built on human-behavioral-modeling and social-intelligence technology for content and marketing.
Anaxi Labs
Pittsburgh, Pennsylvania physical-AI data infrastructure startup founded in 2024 by Carnegie Mellon alumni; runs the WorldSight data-contribution platform.
Generalist AI
Frontier AI research company building embodied foundation models for general-purpose robots, with GEN-1 achieving 99% task success rates.
Void Robotics
Marathon, Florida startup offering sub-$10K autonomous material transport robots at $1/hr; Voidwalking platform moves 50 lbs of goods using GPS navigation.
Fixtureworks
US manufacturer and supplier of modular workholding, fixturing components and standard machine elements for machining, inspection and automation.
Shanghai Artificial Intelligence Laboratory 上海人工智能实验室
Shanghai-based national AI research institute founded 2020; develops the InternLM model family and the Intern-Robotics embodied-AI engine.
Human Archive
Data lab capturing aligned multimodal human-interaction data to train robotics and embodied-AI foundation models.
Fatfish Technology Pte Ltd
Singapore's FatFish AI builds Visual Action Intelligence software, led by its CAMP One engine, for cross-brand robot perception and action.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.