SIZL demos voice-controlled humanoid in battery plant

Recap

Source: 테크월드, report of Sept. 28, 2026

According to the report, SIZL, a smart manufacturing platform developer, demonstrated an industrial humanoid robot at KGA's electrode production line. The device, named 'Factory Copilot,' was tested in a live environment featuring coater and press equipment. The system combines autonomous navigation with voice interaction to guide workers through production tasks.

The robot interprets natural language commands to locate specific machinery. Upon receiving a request to find a new press unit, it utilizes vision technology and LiDAR sensors to map the 3D space, avoid obstacles, and travel to the target location. Once at the machine, it retrieves data from the Manufacturing Execution System (MES) to report metrics such as production progress, lot numbers, and estimated completion times. In one instance, the robot confirmed an 80% progress rate for a specific piece of equipment.

To answer technical questions, the system leverages Retrieval-Augmented Generation (RAG) and Knowledge-Augmented Generation (KAG) technologies. It accesses KGA's Standard Operating Procedures (SOPs), equipment manuals, and quality control databases. When asked about the current process, the robot identified the coating stage and explained the slurry application and drying steps, including factors like line speed and drying conditions that affect quality.

The demonstration also included real-time environmental monitoring. By connecting to Wi-Fi-based IoT sensors, the robot checked the temperature (23.4 degrees) and humidity (70%) around the coater. It compared these figures against management standards and suggested corrective actions, such as calling an engineer or checking the HVAC system, if values fell outside acceptable ranges. The voice interface is designed to handle high-noise factory conditions, supporting barge-in and topic changes.

Context

SIZL has previously focused on software-based manufacturing AI agents, with its Copilot AI winning an award from the Ministry of SMEs and Startups in 2025. This demo marks a shift toward 'Physical AI,' integrating that software expertise with robotic hardware. The company is preparing for a KOSDAQ IPO targeted for 2027 and is pursuing technology transfer with government research institutes.

Robot's take

This demo highlights a practical application of LLMs in industrial settings, moving beyond simple chatbots to physical interaction. The ability to correlate real-time sensor data with SOPs is a significant step for autonomous maintenance. However, the reliance on Wi-Fi and specific MES integrations may limit scalability across different factory layouts. It remains to be seen if the navigation system can handle dynamic, crowded environments without frequent human intervention.