Keymakr launches egocentric and robotics training data solutions for Physical AI systems

New York, NY - [May 19, 2026] - Keymakr, a global provider of high-quality AI training data solutions, launches a new suite of tools and services focused on egocentric data for Physical AI and robotics systems.

Robotics and embodied AI are facing a new bottleneck: robots need data from their own perspective - enriched, structured, and aligned with real-world actions. Keymakr's ready-made datasets are designed to help the industry build high-quality perception, planning, and action models.

Building data from a human-first perspective
Egocentric data captures the world from a human perspective using head-mounted cameras (including FPV setups), body sensors, and wearable devices. These setups record a person's environment, motion, intent, and decision-making process while performing a task.

Keymakr brings 10 years of experience building multimodal datasets into this emerging market. The company uses precise temporal alignment between sensor data, actions, and environmental changes to build state-of-the-art datasets available off the shelf.

"Egocentric data helps AI systems learn how humans approach and manipulate objects, how tools are used, how tasks unfold step by step, and how hand-eye coordination operates in dynamic environments," said Inna Nomerovska, CMO at Keymakr. "To support downstream model training, we offer fully enriched datasets along with a process of creating unique data from scratch with our scalable process. At this point, it's practically Hollywood for AI."

Capturing actions, intent, and interaction
To reinforce the Physical AI division, Keymakr has also introduced new capabilities in its Keylabs.ai platform. These features are specifically designed for egocentric and multimodal robotics data, providing the infrastructure to efficiently manage complex sensor inputs, orchestrate sequence-driven annotation workflows, and ensure high-quality labeling of actions and interactions across dynamic scenarios.

Overall, the new Keymakr Physical AI suit develops datasets that include:
- Multimodal egocentric RGB video streams with synchronized sensor data, including IMU, LiDAR, RGB-D (depth), and hand-joint tracking
- Temporal alignment across modalities and action sequences
- Action labels, activity annotations, and event-based temporal/action annotations
- Object interaction and manipulation tracking
- Hand-joint tracking for fine-grained motion understanding
- Semantic text description layers to connect perception with language-based models
- Vision-language-action (VLA) and multimodal robotics training datasets

In the near term, Keymakr plans to expand into robotic-arm teleoperation data collection, enabling the capture of demonstration datasets on physical robotic systems (minimum 6 DoF). These datasets include joint states, control signals, and action labels, often synchronized with wrist-mounted and third-person camera feeds.

"We emphasize diversity across robotic platforms, environments, and task types to improve generalization of trained models," added Inna Nomerovska. "Now companies are focusing on specific, repeatable tasks performed in real environments. Every dataset should be tied to a real-world workflow and can be adapted to a specific business need. This is where Physical AI is heading, and data is the key enabler."

With the new service launch, Keymakr expands its capabilities from traditional perception datasets to fully structured, multimodal action datasets, enabling companies developing robotics and Physical AI systems based on licensing-compliant, production-ready datasets.

Keymakr
99 Wall Street #423, New York City NY
10005
Olena Butenko PR manager Keylabs.ai/Keymakr pr@keymakr.com

Keymakr is a global provider of AI training data solutions, specializing in data annotation, dataset creation, validation, and data operations for machine learning systems. The company supports organizations building computer vision, multimodal, Physical AI, robotics, and LLM-based agent systems, delivering high-quality datasets through a human-in-the-loop approach and proprietary tools developed within the Keylabs platform.

This release was published on openPR.