A 40-degree-of-freedom humanoid robot with an omnidirectional base, humanoid arms, pure-vision sensing, and optional dexterous hands or two-finger grippers for real-world tasks.
Primary use case: Household, industrial, and commercial scenarios.
DOF
40
Degrees of Freedom
40
Locomotion
wheeled
Hand Type
dexterous
Autonomy Level
unknown
AI Capabilities
vision,planning,manipulation,learning
Onboard Compute
Dual-PC distributed architecture with 1000 Hz real-time joint control and an EtherCAT industrial ring network
Availability
unknown
Primary Use Case
Household, industrial, and commercial scenarios.
Chinese startup DexForce trains humanoid robots using virtual data to slash costs
Shenzhen-based startup DexForce is training humanoid robots using synthetic data generated in virtual environments rather than costly real-world demonstrations, aiming to speed progress toward 'physical artificial general intelligence' and lower training costs. The roughly 500-employee company develops software that simulates physical environments and object interactions, and its systems are used by manufacturers including Toyota, Samsung, Panasonic, Foxconn, Haier and Hisense. Founder and CEO Jia Kui said achieving general-purpose robotics from real-world data alone would require 10 trillion samples, while DexForce previously raised $147 million in a Series B at a valuation above 10 billion yuan (about $1.47 billion).
DexForce is using synthetic data to train humanoid robots in order to reduce costs. The headline-sourced snippet provides no further details about the company, its technology, or the specific robots involved.