LG CNS said on March 10 it made a strategic investment in U.S. robot company Dexmate to strengthen humanoid robot hardware competitiveness optimised for industrial sites.

The investment was made through LG Technology Ventures, LG's corporate venture capital arm. The company said Dexmate, based in Silicon Valley, makes humanoid robots that global robot brain developers have adopted as standard research hardware. Dexmate's robots keep human-like task performance but use wheels instead of legs for stable long-duration work.

They consist of a wheel-based lower body, two arms specialised for high-speed work and a head that detects the surrounding environment with vision sensors. They can perform precise bimanual cooperative tasks based on more than 36 degrees of freedom. The arms support a payload of about 15 kg and can work for more than 20 hours on a single charge. With a more stable lower-body structure than bipedal humanoids, they are easier to apply at industrial sites such as logistics centres and manufacturing plants.

LG CNS stressed that the investment allows it to secure a wheel-type humanoid in addition to bipedal humanoids and quadruped robots, building a lineup of various types of robot hardware. Based on this, it plans to provide a "full-stack RX service" that combines as a package three key elements for humanoid commercialisation: hardware, a robot foundation model (Robot Foundation Model, RFM) and an operations and training platform.

LG CNS is developing its own platform for robot operations and training and continues to expand partnerships and investments for its robot business. In June last year it invested in U.S. robot brain developer Skild AI and formed a partnership to jointly advance industry-tailored RFMs. It is also running multiple proof-of-concept (PoC) projects for robots trained on data from logistics, distribution and manufacturing sites.

Jun-ho Lee (이준호), head of LG CNS' Smart Logistics & City Division, said the investment was a strategic move by LG CNS to enable large-scale robot operations by organically combining robot hardware, RFM and a platform and to rapidly spread them across industrial sites. He said the company will demonstrate a humanoid robot business model that can be applied immediately at real sites, beyond technology verification, to lead the era of physical AI.