DexForce raises $147M to train humanoid robots in simulation

DexForce, a Chinese startup training humanoid robots in virtual simulations, has raised $147 million, according to reports. The capital will fund development of synthetic data pipelines that generate millions of simulated task scenarios—picking, placing, and navigating cluttered spaces—without requiring physical trials. DexForce claims the approach significantly reduces training costs compared with real-world data collection, though it has not disclosed error rates when robots trained in simulation encounter unmodeled conditions.

The round follows recent funding announcements in the robotics training space, including a $60 million raise by Mecka, a U.S. startup building human-motion datasets, and $32 million secured by simulation software provider Antioch in September. Earlier coverage of DexForce highlighted its valuation milestone, though the company’s focus appears to be expanding beyond initial demonstrations.

Competitors like Agile Robots are pursuing alternative methods, such as training industrial arms on video of human workers. DexForce’s approach relies on synthetic data to scale faster, but industry observers note that bridging the “sim-to-real” gap remains a challenge for robotics teams. The next milestone for DexForce may involve deploying its first commercial pilots in the coming years.

Sources: ijr.com