UBTECH has launched Thinker-Sim, an embodied AI simulation platform designed to provide a comprehensive technical framework for digital scene reconstruction, large-scale data synthesis, and robot capability assessment. The system aims to bridge the gap between real-world environments and virtual testing.
The platform utilizes the proprietary Thinker-GS high-fidelity reconstruction algorithm and SimReady asset generation technology. This allows for the efficient conversion of physical environments, such as factory production lines, into 3D digital models. Users can integrate various assets, including robots, equipment, and personnel, to suit specific training and testing requirements.
Thinker-Sim supports diverse applications, including commercial services, by generating photorealistic digital twins. Key technical features include text-to-image 3D asset generation, low-poly topology optimization, visual material creation, and automated collision configuration. These tools balance visual quality with physical simulation accuracy.
To streamline workflows, the platform includes standardized simulation task templates for training and data synthesis. Users can initiate tasks through parameter configuration without building environments from scratch. The system also supports real-time streaming of simulation visuals, task monitoring, and performance feedback.
