RLWRLD Expands Physical AI Push for Industrial Robots
The physical AI developer integrates human‑motion and tactile data with CJ Logistics on technology for managing diverse warehouse items.

SEOUL — Physical AI developer RLWRLD is building a next‑generation robotics foundation model to enhance five‑finger manipulation, as the company works with CJ Logistics to test robot AI in warehouse processes.
CEO Junghee Ryu presented RLDX‑2 strategy at Humanoid Summit Seoul on Sept. 22, according to Edaily. The US’s San Francisco-based company is developing a five‑finger hand with over 20 degrees of freedom.
Ryu cited CJ Logistics operations as an example of the problem RLWRLD wants to address, saying warehouse conveyor speeds exceed those typically encountered in laboratory settings and that robots may need to handle more than 200 types of goods. The report does not establish that RLDX-2 is already processing the goods in normal operations.
The companies' relationship is more concrete than the RLDX-2 project alone. CJ Logistics and RLWRLD entered a partnership this month to jointly develop a robotics foundation model for logistics and will pilot selected warehouse processes. A separate Seoul Economic Daily reported they will test core robot movements before expanding.
RLWRLD's planned RLDX-2 data pipeline is designed to capture human movement directly rather than relying only on teleoperation or first-person video. Ryu described a “4D+” approach combining three-dimensional human-motion data over time with physical signals including touch. External cameras would supplement wearable cameras, while the company is developing thin gloves to collect tactile data.
RLDX-2 builds on RLDX-1, released earlier this year. The existing model integrates three-dimensional human movement data across time with physical signals like touch. Its technical report describes RLDX-1 as a general-purpose robotic policy for dexterous manipulation and documents its performance across simulated environments and real-world applications.
For RLDX-2, RLWRLD is also developing what it calls a “4D+ World Action Model” that would incorporate spatial and physical information when generating robot actions. Ryu said the company's longer-term architecture would use multiple specialized models rather than rely on a single vision-language-action model for everything.
CJ Logistics already has two dual-arm humanoid robots working on a packaging line at its Olive Young logistics center in Yongin, according to the logistics company and The Korea Times. Those robots currently place cushioning material into boxes. CJ Logistics identifies Robotis as its hardware partner, Aidin Robotics for robot-hand technology and RLWRLD for robotics foundation models.
Neither Edaily nor other sources give a release date for RLDX-2, or establishes that the new model is operating commercially.
More articles on this topic
31 articles
ReportOctopus-Inspired Muscle Design Simplifies Dexterous Soft Robots
Report5 Jobs Robots Could Take Over by 2035 — And 5 They’ll Struggle to Replace
ReportBoston Dynamics Sacrifices Atlas’ Fifth Finger for Efficiency
ReportNvidia Develops Safety Controls as AI Advances Into Robotics
ReportWhy aren't robots in our homes yet?
ReportFlourish Opens Preorders for $3,555 Trainable Home Robot
ReportNauticus Robotics Seeks Up to $50M in Strategic Investment
ReportSharpa Unveils Three Tactile Robotics Systems at IROS 2026
ReportHow Do Robots See the World?
Report6 Unusual Jobs Done By Robots
ReportAGIBOT’s 300-Robot Theme Park Moves From Plan to Operation
ReportRealset AI, Flatkey Raise $10M to Expand Real-World Robot Training DataBUSINESS NEWS WEEKLY LETTER
The Weekly Letter for Robotics Professionals, Summarizing the Most Important Industry Moves, Launches, Deals and Signals.