휴먼 어드밴스드 로보틱스 플랫폼 (HARP)

Teaching existing robots has been inefficient, relying on programming or slow motion capture.

Even when trying to transfer a person's delicate skills to a robot, commercial joysticks and motion capture devices have low resolution and response speeds, causing them to lose fine movements. In addition, the lack of a standardized training environment due to different control methods for robots from various manufacturers.

##

HARP's Inverse 3X haptic input device provides 1000 times more precise hand motion capture performance at 32kHz, recording human's delicate manipulations directly to the robot [13].

Based on OpenXR, more than 30 types of heterogeneous robots can be controlled on a single platform, allowing a person to demonstrate once and transfer the technology to various robots [14].

It also accelerates robot learning data collection and reinforcement learning application by more than 10 times in conjunction with NVIDIA Isaac Sim, Groot workflows, etc. [15], and accumulates and utilizes human skills into a Large Behavioral Model by reproducing recorded trajectories at accelerated (up to 4x) or decelerated speeds [16].

In short, it provides an interface that "translates human hand skills into machine learning" [17].

Robot manufacturers, automation system integrators, and research institutions can adopt HARP to utilize it for robot education and development. Industrially, robot development teams in B2B fields such as manufacturing, logistics, and healthcare are the main customers, and it is also used for R&D purposes in universities and research institutes (B2B).

##

The software platform is available to anyone globally, and the haptic device is also universal, so there are few regional restrictions.

However, as it is a solution specialized for advanced robot development sites, it is not expected to directly expand to general SMEs or the consumer market.

Instead, as a leading technology in the "Physical AI" field, it can have ripple effects throughout the manufacturing industry where human-robot collaboration is becoming increasingly important. Through collaboration cases with Hexagon, there is also the possibility of expanding into other industries such as XR design and remote surgery training.

##

It has won the CES Innovation Award for two consecutive years and is recognized as a unique innovation in the field of physical AI training [18][19].

There is a positive response that it "opens a new horizon for human-robot interaction," and it is well-received for its wide range of applications from XR design to industrial collaborative robots [20].

However, in terms of the market, there is also a view that it may remain in a specific expert community rather than becoming popular in the short term because it is a niche product for education/development.