XYZ to integrate BrainX intelligence into Galaxia robots

| NBNTV = Reporter Jeong Dong-jin | Two robotics startups are joining forces.

According to XYZ and Galaxia on the 24th, the two companies will develop robot applications optimized for the Korean market by embedding XYZ's robot intelligence framework, BrainX, into Galaxia's robot hardware.

As Galaxia's local partner in Korea, XYZ will expand its presence among domestic companies and industrial sites while pursuing the software optimization and technical verification required for real-world work environments.

The two companies plan to expand the domestic application of robots, focusing on specific tasks such as retail space automation, clothing folding, and factory operations. They will focus on optimizing robot hardware to meet the specific requirements of each site and implementing them as applications capable of performing actual tasks.

In particular, data collected from actual sites will be used to enhance robot learning and performance. The process involves analyzing data and performance indicators accumulated during task execution to optimize software, which is then re-applied in the field to verify performance. XYZ plans to improve task execution capabilities and field adaptability through this continuous cycle of learning and verification.

Prior to the agreement, the two companies had been verifying technical feasibility using actual robots.

XYZ successfully implemented functions for learning and performing various manipulation tasks by applying its BrainX framework to Galaxia's R1 Lite robot hardware.

Notably, the clothing folding task using the R1 Lite was recently measured to be more than twice as fast as NVIDIA's GR00T 1.6, drawing significant attention. Following this agreement, the companies plan to expand this technical expertise to various tasks across Korean retail and industrial sites.

Hwang Seong-jae, CEO of XYZ, stated, "We are developing models that can be quickly commercialized in the field through Goal-oriented Models specialized for specific tests rather than general-purpose models. We will apply and expand the intelligence design technology we have accumulated to various robot hardwares."