Maniformer at CoRL 2026: Empowering Global AI Model Teams with Real-World Data
AUSTIN, Texas, Nov. 16, 2026 /PRNewswire/ -- At the 10th Conference on Robot Learning (CoRL 2026), one of the most influential academic conferences at the intersection of robotics and machine learning, Maniformer, showcased its real-world data product ecosystem and global data collection network to researchers and industry professionals from universities, laboratories, and technology companies worldwide.
As robotic foundation models advance, this year's conference underscored industry's growing focus on scalable data acquisition: continuously capturing real-world experiences and turning them into high-quality training data to expand robotic capabilities.
At CoRL: Turning Real-World Data into Training Assets
At CoRL 2026, Maniformer showcased its end-to-end data infrastructure for Physical AI, spanning data collection hardware, an all-in-one data platform, and a data crowdsourcing platform.
The MEgo series of data collection devices captures visual, motion, and interaction data from people operating in real-world environments. ManiGo, the company's crowdsourced data platform, brings together contributors and real-world data collection sites across regions, supporting the continuous collection of high-quality data at scale.. MEgo Engine handles data quality inspection, processing, annotation, evaluation, and delivery, transforming raw captured information into structured data assets ready for model training.
A Research Night Bringing Leading Robotics Researchers Together
Maniformer and AGIBOT jointly hosted a Research Social Night, extending discussions beyond the conference into an open setting featuring talks, panel discussions, and networking.
Featured speakers included Yilun Du, Assistant Professor of Computer Science at Harvard University and head of the Embodied Minds Lab; Ran Tian, Research Scientist at NVIDIA ASPIRE; Bolei Zhou, Associate Professor at the University of California, Los Angeles; Manling Li, Assistant Professor at Northwestern University; and researchers from Google DeepMind Robotics.
Discussions covered topics ranging from autonomous exploration and model efficiency to robotic scaling, failure data, and capability evaluation.
Participants also examined how general-purpose capabilities can be translated into reliable performance on specific tasks through targeted post-training. Failure cases, human interventions, and recovery records collected during real-world deployment must be organized into learnable datasets to identify capability gaps and inform subsequent training.
Human video can provide valuable task-level experience, while real-world interactions add essential action feedback. Effective coordination among high-level reasoning, skill execution programs, and low-level perception and control is necessary to balance generalization, real-time performance, and reliability.
More targeted benchmarks can also help identify failure modes, turning evaluation into a feedback mechanism that connects data acquisition, training optimization, and real-world deployment.
The after-party drew over 60 professors, early-career researchers, industry research leaders, and engineers working on robotic foundation models, world models, and robotics data. Attendees came from Meta, NVIDIA, Google DeepMind, Sharpa, ByteDance, and universities including Princeton, Penn, UC Berkeley, Stanford, Imperial College London, and Duke. The gathering connected frontier AI research, data production, and real-world robotics, fostering exchange between academia and industry.
From Austin to Tokyo: Accelerating a Global Data Collection Network
Maniformer is expanding its global presence through a distributed data collection network. The company also exhibited at NexTech Week 2026 Autumn at Makuhari Messe in Japan, with a focus on applications in manufacturing, logistics, and services.
Maniformer currently operates across more than 20 innovation centers in cities throughout China and has established international network nodes in more than five countries. By connecting data collection partners worldwide, the company is rapidly expanding its collection capacity to meet accelerating industry demand.
Across these markets, Maniformer's mission remains consistent: to make data from around the world accessible and useful for AI, accelerating the arrival of the intelligent agent era.
SOURCE Maniformer




Share this article