Asumekku to offer Arducam’s EgoSync-WEAR wearable full-body data collection kit

Asumekku Co., Ltd. will offer Arducam’s wearable full-body data collection kit, EgoSync-WEAR. The announcement title says the kit records human movements and supports the creation of training data for imitation learning, VLA models and other uses. It also describes the kit as “high-precision,” but no accuracy figures or evaluation conditions are available.

Asumekku describes EgoSync-WEAR as supporting the creation of training data for imitation learning, VLA models and other uses. Its presentation as a wearable kit for recording human movements is relevant to developers considering data acquisition as a separate step in developing AI for robots, apart from computing systems and the robots themselves. However, the recording specifications and basis for the accuracy claim are not available, so its performance compared with existing methods and the conditions for its use cannot be assessed.

EgoSync-WEAR is positioned not as a robot or an inference computer, but as a tool for turning human movements into training data. The announcement title describes wearable full-body data collection and support for creating data for imitation learning and VLA models. The key point, therefore, is that the product is presented as a way to record movements at a stage preceding robot control and execution. However, “high-precision” is the announcer’s description, and no accuracy metrics or measurement conditions are available. The title alone also does not show what movements can be recorded or how the data is formatted for model training. In related coverage, we have reported that Asumekku introduced a camera connection development kit, the development kit supporting up to eight cameras for Jetson Thor, and announced it would handle an edge AI computer for robots. We have also reported the start of sales of the collaborative robot arm myCobot Pro 630. These products and the new kit relate to different stages: image capture and computing, the robot itself, and movement data acquisition. The announcement does not say whether the products connect as an integrated system or whether the data is limited to particular robots. The broader industry implication is that, in developing imitation learning and VLA models, not only computing resources and robots but also methods for capturing real-world movements and converting them into training data can become product offerings. However, because the data acquisition method is not specified, it is not possible to assess how it differs from existing camera-based collection methods or how it handles time synchronization and body-part estimation. Details to look for next include the measurement method and definition of accuracy, recordable movements and data formats, output methods for training environments, price and availability date. Without this information, the conditions for adoption in development settings and which part of the training-data creation process the product could replace cannot be determined.

The product name confirmed in the announcement title is “EgoSync-WEAR.” It is described as a wearable full-body data collection kit for recording human movements and supporting the creation of training data for imitation learning, VLA models and other uses. The available primary-source information does not specify the types or number of sensors, recording method, data format, accuracy measurement conditions, price or start date for availability.