Key takeawayBuilding on its industrial simulation capabilities, Suochen Technology is expanding into physical AI models, synthetic data and robot training toolchains.
How are the two product lines divided?
- Suochen Technology has established two product lines, “Tiangong” and “Kaiwu.” The former covers industrial simulation, automation software, control systems and production management solutions; the latter targets complex physical scenarios and covers physical AI development, model training, agents and synthetic data.
- Building on its existing physical AI development and application platform and smart workshop, the company has also released agents, synthetic data products and a world physics model, targeting intelligent simulation workflows, data generation and deep physical-field simulation, respectively.
What problem does the world physics model seek to solve?
- The company says its “Yingzao · Wanxiang” world physics model is driven by high-precision physics and combines AI perception, generation and prediction capabilities to extend simulations from observable surface environments to deep physical fields.
- Its capabilities include vision-action models; multimodal models for force, touch and sound; and simulations of hidden physical phenomena such as internal material stress, airflow shear and electromagnetic interference.
Which areas does the cooperation with Unitree Technology cover?
- Suochen Technology has signed strategic cooperation agreements with Unitree Technology and others to jointly establish the Qiantang District Unitree New Quality Productive Forces Innovation Experimental Base. The cooperation covers technical training, scenario validation and the development of public service capabilities.
- As part of the base cooperation, Suochen Technology focuses on providing domestically developed physical AI simulation software and supporting toolchains for virtual reconstruction of real-world scenarios, multimodal synthetic data generation and quality validation. It also supports model training, policy transfer and performance evaluation in simulated environments, connecting virtual training with real-world scenario validation.