A small experiment began in late 2024 at Unitree 's headquarters in Hangzhou, China's largest robot corporations. The work involved mounting the Autonomous Driving system from Korean robot startup Neubility onto a Unitree quadruped robot.
Unitree was looking beyond simply selling robots in the Korean market to find a solution partner that could actually deploy them on industrial sites. Neubility was at a point where it sought to expand the Autonomous Driving technology it had built in delivery and patrol robots to quadruped and Humanoid Robot platforms. As their needs aligned, about a year and a half of technology validation and demos followed, and on the 25th the two companies signed a strategic memorandum of understanding (MOU) to jointly develop robot solutions for industrial sites.
The core of this collaboration is attaching Neubility's "Autonomous Driving brain" to Unitree robots. Neubility's Autonomous Driving module "NeuOn" is a kit that can be installed on existing robots to enable autonomous functions. It helps robots perceive their surroundings, plan routes, and avoid obstacles to move on their own. When combined with a control platform that manages multiple robots at once, Unitree 's quadruped and Humanoid Robot platforms can be used for patrol, detection, transport, and special missions.
Neubility knocked on the door first. At the time, Neubility was developing and producing wheeled delivery and patrol robots, but it did not see a need to confine the Autonomous Driving technology it had built to wheeled robots alone. Unitree 's push to target industrial sites with quadruped and Humanoid Robot platforms offered Neubility a new opportunity for collaboration. Places like manufacturing and construction sites or disaster zones, where the ground is uneven or obstacles abound, are difficult to handle with wheeled robots alone. Neubility judged that by mounting an Autonomous Driving "brain" on Unitree robots, it could broaden the application of its own technology.
Unitree also had a hurdle to clear. Building a good robot body and actually operating it at an industrial site are different matters. To deploy robots in factories, security facilities, and construction sites, you must design site-specific routes, safety standards, mission methods, and remote control systems together. What Unitree needed was operational experience to turn hardware into services tailored to industrial sites, and Neubility is a company that has tested this process at multiple domestic sites.
Neubility's robots had amassed real-world driving experience close to "two laps around the Earth." At 142 locations nationwide, 305 wheeled Autonomous Driving robots performed more than 40,000 delivery and patrol services annually. The cumulative driving distance reached 78,497 kilometers, and this operations data is fed back to further enhance Autonomous Driving performance and the control system. The more data that is accumulated, the more refined the site-specific routes, mission methods, and remote response systems become.
Last month at the AWS Summit exhibition at COEX in Gangnam-gu, Seoul, they unveiled to the public for the first time the collaboration model they had been validating. "NewTrek," which mounts Neubility's autonomous navigation module on Unitree 's A2 quadruped robot, was demonstrated autonomously driving along a designated route and linking with the control system. Based on this, the two companies will first develop robot solutions applicable to industrial facilities, construction sites, and security and reconnaissance in the Korean market, and later expand the scope of cooperation to the Asia-Pacific region.
Industry insiders view their collaboration as part of a shift in robot competition toward field validation and reference building. Quadruped robots and humanoids are appearing one after another, but what many robot corporations are desperate for are cases of repeated operation in real-world sites. The industry is watching to see whether this experiment, which begins with a focus on the Korean market, can take root as robot services that work in sites across the real world.