Reportedly: X Square Robot Launches X2Real, an Evolving Simulation Benchmark It Says Tracks Real-Robot Results
Reported by Pandaily · not yet confirmed by the company or a second independent outlet. We update this page when it is.
X Square Robot launched X2Real, a simulation benchmark that calibrates simulated robots against real hardware, spans 44 tasks in 10 categories, and uses AI agents to write new tests. The company reports a 0.96 sim-to-real success correlation.
China context
- Original name
- 自变量机器人
- Outside China
- Not stated in the sources yet
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Robotics developers outside China should verify whether X2Real is publicly accessible and independently test its sim-to-real correlation before adopting it for their own evaluation pipelines.
- For investors
- Investors should seek independent validation of the 0.96 correlation and assess whether X2Real becomes a widely adopted standard, as that would indicate X Square Robot's influence in the embodied AI tooling market.
Translated from Chinese. Quotes and facts link to the original sources.
X Square Robot has introduced X2Real, a benchmark designed to evaluate embodied AI by aligning simulated robot performance with real-world results. The benchmark covers 44 tasks across 10 categories and employs AI agents to generate new test scenarios. According to the company, X2Real achieves a 0.96 correlation between simulated and real-robot success rates.
X2Real's reported 0.96 sim-to-real correlation suggests a strong alignment between simulation and physical performance, but this figure is company-reported and not independently verified. The use of AI agents to generate new tasks indicates an evolving benchmark that could adapt to new capabilities, though the specific generation methods and validation processes are not detailed in the evidence.
Robotics developers outside China gain a potential tool for evaluating sim-to-real transfer, but its practical value depends on independent validation of the 0.96 correlation and access to the benchmark. If the benchmark is not publicly available, its impact on the broader robotics community remains limited.
For robotics companies, a reliable sim-to-real benchmark could reduce the cost and time of physical testing by identifying simulation gaps early. However, the business value is contingent on the benchmark's accessibility and proven accuracy beyond the company's own reporting.
The next verifiable signal is whether X2Real is released publicly or remains internal, and whether independent researchers replicate the reported correlation. Without such confirmation, the benchmark's influence on embodied AI development cannot be assessed.