Gasgoo Munich-At the 2026 World Artificial Intelligence Conference (WAIC 2026), GenRobot AI took the stage with a complete end-to-end Human Data solution. Built on DAS acquisition hardware, the DFM data foundation model, and the ADP scalable data pipeline, the system establishes a massive, diverse data supply chain. This chain is centered on full-modal perception capabilities for the head, hands, and body.
Leveraging a complete industrial closed-loop capability, GenRobot AI is focused on building full-stack data acquisition for embodied AI—covering the head, hands, and body. Through a suite of solutions comprising Ego, Dex dexterous gloves, and full-body Mesh motion capture, the company systematically captures human first-person environmental perception. It also records fine hand manipulation, fingertip tactile contact, and continuous full-body movement trajectories. The result is a standardized Human Data foundation—high-quality, multi-modal, and ready for model training. It provides stable, comprehensive underlying data support for various humanoid robots and embodied AI models.
Image Source: GenRobot AI
Specifically, Ego is designed to replicate the observational logic of the human eye. By fully recording environmental visuals and perspective shifts within task scenarios, it authentically reconstructs the human process of observing the environment and judging task intent. This serves as the perceptual entry point for Human Data.
The Dex dexterous gloves focus on fine human hand interaction. They comprehensively replicate multi-dimensional joint movements, precisely capturing delicate operations such as grasping, twisting, and pressing. By simultaneously recording tactile contact and force variations, the gloves preserve every detail of fine human manipulation—acting as the core operational data source for Human Data.
For full-body movement, the company relies on its continuously iterated full-body Mesh solution. This focuses on collecting human posture trajectories, limb linkage, and spatial balance data—filling the gaps left by single-perspective or single-hand operations. It perfects Human Data’s ability to fully express overall human behavior and coherent movement.
Paired with its proprietary DFM data foundation model, GenRobot AI has further achieved the alignment, reconstruction, and standardization of multi-source raw data. The system transforms raw captured material into high-precision, time-synchronized structured data ready for model training. This offers a stable, standardized, and multi-modal high-quality data foundation for various humanoid robots and large embodied AI models.
This means GenRobot AI is not limited to developing single-point devices. Instead, by combining a four-dimensional acquisition system covering the head, hands, body, and full senses with DFM data processing capabilities, it is building an underlying data ecosystem. This ecosystem is adapted to general embodied AI development. The company provides high-quality, mass-producible, full-dimensional core data support for humanoid robots, embodied large models, and dexterous manipulation algorithms.
Currently, GenRobot AI’s specialized embodied AI data production line covers over 3,000 collection users and more than 10,000 real-world scenarios—including homes, factories, commercial sites, logistics centers, laboratories, and medical facilities. To date, the company has accumulated over 1 million hours of real-world scenario data assets, encompassing more than 2,000 human daily operational skills.
