Lumos to give away 100 robots to global builders, cites access as bottleneck

Lumos Robotics embodied AI
  • Chinese startup LUMOS is giving away 100 humanoid robots to accelerate embodied AI development
  • The program launches alongside a more agile NIX robot built for advanced motion and interaction

Lumos Robotics (鹿明机器人), a Suzhou-based embodied AI and robotics startup, said it will provide 100 of its NIX humanoid robots free of charge to selected universities, robotics laboratories, developers and creative technologists worldwide under a new initiative aimed at expanding access to humanoid robotics platforms.

The program, known as Project EDGE — the Lumos NIX 100 Builders Program, will provide recipients each with a NIX robot, access to the company’s software development kit and technical support from its engineering team.

Lumos said participants will not be required to pay licensing fees, provide equity or make upfront purchases.

The company said the initiative stems from a view that access to humanoid robots remains a major constraint on embodied AI development.

Research platforms often cost between $50,000 and $100,000, limiting participation by universities, independent developers and smaller research teams.

Removing the cost barrier

Rather than focusing solely on internal development, Lumos said it hopes to encourage a broader range of experimentation in areas such as human-robot interaction, motion generation and real-world embodied AI applications.

“We believe that by removing the cost barrier, we can unlock the kind of experimentation that the field has been missing,” said Yu Chao, founder and CEO of Lumos Robotics.

The announcement comes alongside an upgraded version of the company’s compact humanoid robot NIX.

According to Lumos, the latest model features improvements in dynamic motion control, whole-body coordination and the execution of complex movements.

Advanced locomotion

In a recently released demonstration video, NIX performs rapid dance sequences, agile footwork, backflips, side handsprings, one-arm handstands and windmills.

The company said the demonstrations reflect advances in motion control, embodied AI and human-robot interaction.

The robot’s movement capabilities are powered by a control system trained through a combination of motion-capture data, simulation and reinforcement learning.

Rather than simply replaying recorded human movements, the system is designed to adapt actions to the robot’s own physical structure and mechanical constraints.

Lumos said it generates training data using animation-derived reference trajectories, simulation optimization and manually designed motion sequences.

The objective is to enable the robot to maintain balance, smoothness and energy efficiency while executing dynamic movements.

Before deployment, NIX undergoes large-scale parallel simulation training that exposes it to varying landing conditions, contact patterns and external disturbances.

During operation, the robot continuously monitors joint states, body orientation, velocity and ground-contact conditions, allowing it to adjust to changes in friction, posture and external forces in real time.

All photos courtesy of Lumos Robotics

The underlying hardware includes Lumos’ proprietary P-60 joint module, which integrates motors, sensors and control electronics into a compact package.

Each module weighs 570 grams, measures 64 millimeters in diameter and can reach rotational speeds of up to 160 revolutions per minute, according to the company.

Participants in Project EDGE will also be able to submit custom motion commands and movement fragments through standardized interfaces.

The company said submitted motions will undergo simulation testing, dynamics validation and safety screening before being converted into executable robot actions.

A larger ecosystem

Lumos said the initiative is intended to create a larger ecosystem for humanoid motion development by allowing external developers to test and deploy new movement concepts on a production-grade robot platform.

The company has previously drawn attention from the global robotics community. In January 2025, Brett Adcock, founder and CEO of Figure AI, commented publicly on Lumos’ earlier LUS1 humanoid robot after the company demonstrated its fall-recovery capabilities.

“Chinese startup Lumos Robotics shared progress on LUS1 humanoid, which can now face hurdles, fall and stand back up again,” Adcock wrote on social media at the time.

In a later post, he noted that the robot was able to recover from a fall in roughly one second, describing the system as being designed for operation in unstructured environments.

Interested developers, researchers, and robotics teams can apply to join the Project EDGE Builders Program at https://www.lumosbot.tech/project-edge/index.html.