South Korea-based LG CNS has announced a partnership with US robotics startup Skild AI, marking its entry into the industrial AI humanoid robot market.

The agreement includes an investment through LG Technology Ventures, the corporate venture capital arm of LG.

LG CNS plans to use Skild AI’s RFM technology to create AI humanoid robots tailored for sectors including manufacturing, logistics, and urban services.

These robots are designed to handle equipment monitoring, product assembly, hazardous material handling, and logistics operations.

The partnership aims to streamline robot operations by minimizing the need for task-specific model development. LG CNS intends to integrate these AI humanoid robots into services for smart factories, smart logistics, and smart cities.

Skild AI, co-founded by Carnegie Mellon University professors Deepak Pathak and Abhinav Gupta, focuses on developing robot foundation models (RFM).

This technology allows robots to learn and perform tasks autonomously using data from various sources, such as images, text, audio, and video.

🔗 Source: The Korea Times

🧠 Food for thought

1️⃣ Robot foundation models signal a fundamental shift in robotics development approach

LG CNS’s partnership with Skild AI highlights how robot foundation models (RFMs) are revolutionizing industrial robotics by enabling generalized learning rather than task-specific programming.

Historically, industrial robots required extensive programming for each specific task, making deployment costly and time-consuming across different applications.

Traditional robotics development followed a labor-intensive approach where engineers had to explicitly program each movement and response, limiting adaptability to new environments.

The shift to foundation models parallels the transformation we’ve seen in AI development, where pre-trained models like GPT revolutionized capabilities through massive data training rather than rule-based programming.

Skild AI’s approach, led by Carnegie Mellon professors, allows robots to learn from diverse data sources including video demonstrations, dramatically reducing implementation time while increasing flexibility across manufacturing and logistics applications.

This represents a fundamental evolution from the first industrial robot Unimate (1961), which could only perform pre-programmed repetitive tasks, to today’s AI systems that can autonomously learn and adapt to new environments.