Emerging Challenges in Embodied Intelligence: Data Quality vs. Quantity for Robots
- Published
- Source
- leaderobot.com
At the WRC, insights revealed that the next phase of embodied intelligence is facing a paradox: the scarcity lies not in data volume but in the physical world representation within that data. Over the past year, numerous training centers have emerged across China, with over 90 expected to be operational by mid-2026, generating millions of data points. However, only a fraction of this data is applicable to real-world tasks, indicating a potential issue as the industry approaches commercialization.
This situation is critical because, unlike language models, robots cannot rely on the internet for training; they must navigate real-world complexities such as friction and collisions. The founder of Orbbec, Dr. Huang Yuanhao, emphasized that while digital intelligence thrives on vast amounts of text data, embodied intelligence requires experiential learning from human operators, which is currently lacking.
Recent research indicates a shift in focus from merely increasing demonstration data to enhancing 3D spatial understanding for foundational robot training. Innovations like FreeTacMan's wearable devices are also emerging, allowing humans to gather rich interaction data directly. As the industry explores these new avenues, the question remains whether the bottleneck in robot training is due to insufficient data or the high costs and limitations of current data production methods.
Original report
WRC Insights: What Real-World Experience Do Robots Lack Today?
leaderobot.com · LeaderobotThis 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
IFA 2027 – Berlin, Sep 3–7, 2027 RobotToday · 3 days ago
- 9
- 10
- 11
- 12
Related Suppliers
Orbbec Inc. 奥比中光科技集团股份有限公司
China
Chinese 3D vision sensor manufacturer founded in 2013, publicly listed on Shanghai STAR Market, serving robotics, logistics, and healthtech globally.
Xamla Robotic Solutions (Provisio) GmbH
Germany
ROS-based robot programming IDE (ROSVITA) and 3D sensor systems; internal startup of PROVISIO GmbH, Munster, Germany.
Lightwheel
United States
Physical AI infrastructure company providing SimReady assets, egocentric data, and evaluation platforms to train and deploy robots at scale.
University of Michigan, Mobile Robotics Lab
United States
Ann Arbor, Michigan university lab founded 2006 by Prof. Ryan Eustice; studies SLAM and perception for underwater, ground and automotive robots.
Lumos Robotics 鹿鸣机器人
China
Shenzhen embodied-AI firm (est. 2024) making LUS biped and MOS wheeled dual-arm robots plus FastUMI robot-free data capture; A-round led by Mitsubishi Electric.
Physical Intelligence
United States
San Francisco AI robotics company founded 2024; its pi0 vision-language-action models direct many robot types, backed by billions in venture funding.
Physical Sciences Inc. (InstantEye Robotics)
United States
Massachusetts R&D firm founded 1973; InstantEye division makes hand-launched tactical quadrotor sUAS used by U.S. military and first responders.
Noitom Robotics Technology (Shenzhen) Co., Ltd. 诺亦腾机器人
China
Shenzhen embodied-AI data company building motion-capture and teleoperation datasets for humanoid robots.
GeoComm
United States
St. Cloud, Minnesota GIS specialist founded in 1995 that builds NG9-1-1 location-intelligence infrastructure now linked to drone-as-first-responder dispatch.
Shenzhen Stella-Robot Technology Co., Ltd. 深圳星际光年机器人技术有限公司
China
Shenzhen embodied intelligence startup founded in 2024, developing full-stack dexterous hand and joint module technology for humanoid robot manipulation.
GENISOM AI 智深科技
China
Full-stack embodied-AI company building industry-grade quadruped robots, CHAMP joint actuators and the open-source MATRiX simulation platform.
Sequor Robotics
Korea, Republic of (South)
Seoul, South Korea robotics software firm founded December 2022; builds 3D vision and simulation tools that let robots learn from real spaces.
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.
