- Embodied AI platform targets a key bottleneck in industrial robot deployment
- Lumos says enterprises can develop, test and deploy new robot skills without building a full AI engineering stack
Lumos Robotics (鹿明机器人), a Chinese humanoid robotics and embodied AI startup, has launched NexCore, a platform aimed at shortening the time it takes to teach industrial robots new skills from weeks to days.
The launch, announced on September 16, targets a problem that often emerges after a robot reaches the factory floor: getting it to perform a specific task reliably in a real production environment.
“The key to scaling embodied AI is making the cost of acquiring a new skill low enough,” said Yu Chao (喻超), founder and CEO of Lumos Robotics. “Our goal is to make robotic skills producible at scale.”
Deploying a robot can require more than programming. Companies may need to collect and process data, develop algorithms, train models, deploy them to the robot, validate performance and troubleshoot failures on site.
A change in product, workpiece or production environment can trigger another round of development and testing.
Engineering-intensive
For companies without their own robotics AI teams, this can turn what looks like a hardware deployment into a lengthy engineering project.
According to Suzhou-based Lumos, NexCore brings these steps into a single workflow, covering task definition, data ingestion, model training, skill creation, evaluation, deployment and continuous learning.
Users can describe a task in natural language and connect real-world, simulation and video data.
They can then train candidate skills, evaluate their performance and deploy validated capabilities to compatible robots.
The idea is to shift robot deployment from a series of one-off engineering projects toward a more repeatable process for producing and improving skills.
From deployment to iteration
The need for such a process becomes more pronounced as industrial robots move into less standardized environments.
A robot working on one production line may encounter different products, tools or operating conditions on another, requiring its skills to be adapted or retrained.
NexCore is designed to keep those changes within the same development workflow. Once a skill has been developed and validated, it can be packaged for reuse across compatible robots and similar tasks.
Data generated during production can then be fed back into the system to help refine the skill.
Reusable skills
That means experience from one deployment can potentially be carried over into the next, rather than requiring customers to start from scratch each time.
The approach also reflects a broader shift in industrial robotics.
As hardware becomes more capable, the challenge is increasingly less about putting a robot on the factory floor and more about giving it the skills to handle changing real-world tasks reliably.
*This is a press release provided by Lumos Robotics. Please visit https://www.lumosbot.tech/ for further information.
Header image courtesy of Lumos Robotics
