en.Wedoany.com Reported - On June 30, Infinite Force officially launched the DataGrid end-to-end embodied intelligence training system, which consists of a data collection system, a data processing platform, and a hardware base. The goal is to connect data production, data governance, model training, and hardware verification required for robot training into a complete pipeline. Founded in 2023, Infinite Force publicly states that it focuses on embodied brain and robot development, with R&D networks in Hangzhou, Shenzhen, and Beijing, covering business areas including embodied brain, full-stack AI infrastructure, and robot products.
DataGrid is not a single collection tool but an engineering system built around embodied intelligence training.
The current challenge in embodied intelligence has shifted from "whether a demonstration action can be performed" to "whether high-quality data can be continuously obtained and transformed into reusable capabilities." When robots enter real-world scenarios, they encounter issues such as lighting changes, object occlusion, spatial differences, task interruptions, human-robot coexistence, and path variations. Single teleoperation sessions, small sample sizes, and closed experimental environments struggle to cover these variables. Industry research also indicates that robot training data typically requires multi-source heterogeneous information such as vision, touch, force, motion trajectories, and body states, which cannot be easily obtained from the internet on a large scale like data for large language models.
Infinite Force has previously placed the data loop at the center of its technical system. Public information shows that its full-stack AI infrastructure covers data collection, simulation generation, model training, and deployment evaluation. CTO Wang Yizhou has introduced engineering loops, data loops, and training platform methods accumulated during the autonomous driving era into the embodied intelligence R&D system.
The three components of DataGrid announced this time correspond to three key nodes in the embodied intelligence training pipeline. The data collection system is responsible for obtaining samples from real tasks, robot operations, and scene interactions. The data processing platform handles cleaning, annotation, segmentation, quality inspection, format conversion, and training task organization. The hardware base is responsible for action execution, perception verification, and task reproduction after the model returns to the robot body. When combined, training no longer remains a discrete process of "collecting a batch of data, training a model once" but forms an iterative, traceable, and reusable production system.
The Suzhou municipal government previously disclosed that the Suzhou Industrial Park and Infinite Force jointly established the "Embodied Intelligence · Brain Training Center," forming a complete technical loop from data collection, processing, and training to simulation verification. This aligns with the direction of DataGrid's release and also shows that Infinite Force is extending its embodied brain R&D from the model layer to the training infrastructure layer.
From an industry rhythm perspective, embodied intelligence companies are competing not only for robot hardware but also for data, training platforms, and scenario loops. Hardware forms can iterate quickly, but whether robots can enter complex environments such as supermarkets, warehouses, manufacturing, services, and homes depends on whether the model has seen enough tasks, objects, spaces, and anomalies. The value of DataGrid lies in integrating these scattered variables into a unified training process, forming a continuous loop of data collection, data processing, model training, and hardware verification, reducing the cost of re-customization each time a new scenario is entered.
Infinite Force's release of DataGrid indicates that its embodied intelligence roadmap is expanding from "embodied brain" to "intelligent training system." If the scale of data collection, hardware adaptation range, training efficiency, and real-world scenario verification results continue to be disclosed, DataGrid will become an important entry point for observing Infinite Force's commercial capabilities in embodied intelligence.
