디클록브레인 PriviSphere 로보틱스

For AI robots to operate in factories, hospitals, and public spaces, they need to recognize the images of people around them, which raises significant concerns about privacy infringement. Furthermore, individual robots collect information independently without sharing it, resulting in low collaboration efficiency and difficulties in central control. Existing robot vision systems mostly send data to the cloud for AI processing, which leads to intelligence degradation and security risks when the network is disconnected.

The PriviSphere Robotics platform anonymizes personal video data at the local edge (deleting personal faces/identifying information from real-time video) and protects privacy by sharing only the model parameters learned by each robot through federated learning, without sharing the original data. This allows multiple robots to integrate their learning results and perceive space as a team and divide tasks without a central cloud.

For example, multiple robots in a factory combine their camera views to identify the location of people/obstacles without blind spots, and the central control center comprehensively monitors the situation without personal information. In addition, with the AI Personal Information Protection Agent (AipA) and Vision-Language-Action integration (VLA) functions, the robot understands and acts on natural language instructions according to the situation, without leaking personal information.

In summary, it is differentiated from existing standalone robots in that it is a platform that safely connects the eyes and brains of multiple robots into one.

Smart factories, medical institutions, security companies, etc. (B2B/B2G) can adopt this platform to operate multiple robots. For example, in hospitals where patient privacy is important, robots can patrol hospital rooms, and robots can work together in factories while protecting worker privacy.

Robot manufacturers may also adopt this technology in the form of a license to incorporate it into their robots.

It has high global scalability as it can be applied to all robot utilization fields worldwide where personal information protection regulations are being strengthened. Currently, it is being deployed mainly in the Asian market, including Taiwan and Hong Kong, but there is expected to be demand in strict GDPR markets such as Europe.

It can also be extended to fields where multiple devices collaborate, such as smart city control and autonomous vehicle platooning. However, collaboration with individual robot manufacturers and on-site verification will take time, so it is expected to gradually spread from specific industries rather than rapid popularization.

As a robot AI that directly addresses privacy and security issues, it was selected as a CES Honoree and was evaluated as an "essential technology for robot deployment in the future." There is also market expectation that it has opened up new possibilities for companies that have been passive in introducing robots due to data security concerns.

On the other hand, since actual field testing and stability verification are still in the early stages, some experts have mentioned that "the idea is excellent, but concrete performance verification is additionally needed." Overall, positive views are dominant, and many see it as a stage where market verification is still insufficient rather than being undervalued.