CNTXT AI announced an investment plan worth 50 million US dollars (equivalent to 183 million UAE dirhams) aimed at enhancing its operations in the physical AI sector. This investment focuses on developing robotics hardware, building the necessary infrastructure, and establishing specialized facilities for training and testing operations in both the United States and the Middle East region.
Specialized facilities and a vast data network
The company intends to establish advanced technical facilities in the cities of Dubai and Abu Dhabi, specifically designed to simulate real-world work environments and homes, allowing robots to be tested under conditions resembling daily life. To provide the necessary data for these operations, CNTXT AI relies on an operational model that connects robotics developers with a network of 500 partner companies in various sectors such as manufacturing, hospitality, logistics, and healthcare.
The company also benefits from a massive human database comprising over 100,000 contributors in more than 20 countries, who document daily tasks via the KLKT application. This data undergoes automated checks and rigorous human review to ensure its quality before being used to train robotic models.
Advanced motion capture technologies
As part of its quest to provide high-precision data, the company has developed KPTUR, a wearable device (worn on the head or wrist) dedicated to recording precise human movements and hand interactions with objects. This technology aims to provide developers with detailed data that helps robots learn physical tasks requiring high control.
The company is also working on preparing data sets to comply with standard technical frameworks in the field of robotics training, such as LeRobot and RLDS, to ensure their easy integration into machine learning systems.
What changes practically?
This direction represents a solution to one of the biggest obstacles in robotics development, which is the data gap between simulation and reality. Instead of relying solely on virtual environments, CNTXT AI provides data derived from real human movement in actual work environments.
Furthermore, the company seeks to address a technical phenomenon known as "Diversity Collapse", which is a problem that leads to the limited ability of robots to adapt to variables when trained on narrow data. The company has filed a patent application in the United States for a methodology that addresses this phenomenon, which could contribute to producing more flexible robots capable of working in unpredictable environments.