[Edaily Reporter Shin Young-bin] #Cimes Robotics, an AI-based robotics solutions company, is expanding the industrial application of physical AI through its collaboration with NVIDIA.

Cimes Robotics announced on the 9th that it attended NVIDIA’s “Korea AI Ecosystem Reception,” held at the Shilla Hotel in Seoul on the 8th.

The event was hosted by Jensen Huang, CEO of NVIDIA, and attended by approximately 30 representatives from AI startups, including domestic physical AI value chain companies and large language model (LLM) developers.

Cimes Robotics is developing robotic automation solutions that combine 3D vision technology with AI. Based on full-stack AI technology—encompassing AI training, deployment, operation, and verification—the company implements physical AI robot solutions that recognize the shape and location of objects in industrial settings and make decisions and perform actions according to the situation.

The company applies its robotic solutions to areas requiring precision and flexibility, such as real-time sorting and loading of goods in logistics facilities, automotive parts assembly in manufacturing plants, and electronics manufacturing processes. A key feature is its ability to recognize objects and make decisions based on on-site data to perform tasks.

Cimes Robotics extensively utilizes NVIDIA’s hardware and software toolchain to implement its physical AI technology. During the robot AI model training phase, the company operates its own servers equipped with NVIDIA GPUs, including the B200, H100, H200, and A100.

For the stage of applying trained models to the field, the company has built a system that uses NVIDIA Triton Inference Server to analyze multiple 3D data streams simultaneously in real time. This enables the AI to simultaneously process data from multiple vision cameras installed in factories and logistics centers and utilize it for robot control.

Additionally, the company is building an on-premises environment that utilizes on-site GPU clusters rather than relying on cloud servers. This is to ensure fast and stable AI decision-making and robot control in industrial settings without network latency.

NVIDIA technology is also utilized in physical AI and humanoid robot research. By leveraging NVIDIA Isaac Sim, the company has built a digital twin environment that closely resembles an actual factory, where robots undergo repeated verification and training in a virtual environment to ensure they can respond to various situations.

In the industrial AI and robotics sector, the company is implementing Jetson-based edge AI platforms and is also exploring the potential of Jetson Thor, a next-generation platform for humanoid and general-purpose robots.

Lee Seong-ho, CEO of Cimes Robotics, stated, “Based on NVIDIA’s full-stack AI and robotics technology, we will accelerate the on-site application of physical AI and humanoid robots and lead the way in intelligent automation for the manufacturing and logistics industries.”