CJ Logistics deploys humanoid robots at live logistics center

CJ Logistics has deployed humanoid robots in live logistics center operations, moving beyond last year's feasibility tests to place the machines directly on a production line processing real customer orders. The company plans to start with simple, repetitive tasks before gradually expanding the robots' roles to include picking, sorting and inspection.

The company said Thursday it has begun operating two dual-arm humanoid robots at the CJ Olive Young logistics center in Yangji, Yongin, Gyeonggi Province.

The robots' first assignment is inserting cushioning material into boxes along an active packaging line, working alongside human employees on shipments bound for actual delivery.

The deployment marks a step beyond last year's pilot at the Gunpo fulfillment center, which focused on verifying whether humanoid robots could move and function reliably in a logistics environment. This time, the robots have been placed inside a live operational workflow.

CJ Logistics plans to feed data generated by the robots during actual work back into an AI system for further training. The goal is to improve accuracy through repetition and build the robots' ability to adapt when products or working conditions change.

Logistics centers are widely considered harder to automate than factories. Manufacturing plants typically assemble fixed components in a set sequence, whereas logistics centers handle a constant flow of products that vary in size, shape, weight and material — and the items to be processed and the order of tasks shift with each incoming order.

Conventional conveyor systems and fixed-position robots can handle specific tasks quickly, but adding new processes often requires reconfiguring equipment or reworking the workspace.

Humanoid robots offer a different advantage: they can be introduced without major changes to spaces already designed for human workers. Their arms and hands allow them to perform a range of tasks, making them relatively adaptable to logistics centers where products and processes change frequently.

CJ Logistics is investing heavily in the robots' "brain." The company is refining a robot foundation model, or RFM, using data gathered from logistics operations in the field.

The RFM is a physical AI technology that enables a robot to assess its surroundings through cameras and sensors and independently determine what action to take. CJ Logistics plans to train the model on both real-world operational data and simulation data generated in virtual environments, strengthening the robots' ability to handle unfamiliar products and settings.

While the humanoid robots are currently limited to inserting cushioning material, the company intends to expand their role in stages — ultimately assigning them to more complex processes such as picking, sorting, inspection and packaging.

The longer-term vision is to develop each unit into a general-purpose robot capable of moving between multiple processes and handling them in sequence, rather than repeating a single fixed task.

CJ Logistics is also pursuing external partnerships. It is collaborating with Robotis on robot hardware, Eidin Robotics on robotic hands, and RealWorld AI on RFM development. The company is also participating in government research and development projects related to human-shaped robotic hands.

"This deployment is significant because it moves beyond demonstrations of technology or proof-of-concept pilots — humanoid robots are now performing actual roles within live logistics operations," said Kim Jeong-hee, head of the CJ Logistics TES Logistics Technology Research Institute. "We will combine the vast logistics data accumulated on the ground with AI technology to develop humanoid robots that can independently judge and carry out a wide range of tasks, and work toward realizing a future fulfillment model where humans and robots work side by side."