humanoid.guide publishes article covering Sharpa’s “CraftNet”

On November 15, 2026, humanoid.guide published an article about Sharpa’s CraftNet, titled “Welcome, CraftNet!”

This newspaper’s previous coverage of Sharpa has included products on the robot and hand hardware and operation side, as well as examples of learning using tactile sensing. Sharpa unveils three products at IROS 2026 covered its product lineup, while Nikkei Xtech introduces a case of Sharpa using tactile sensing covered an example involving the operation of a five-fingered hand. The publication of humanoid.guide’s article is simply another angle on Sharpa, distinct from that previous coverage. The fact that the article was published does not establish that a product is available, operating in a real-world environment or proven to perform as claimed. In robotics, a key issue is not only the capabilities of robots and hands, but also how real-world information, including contact, can be incorporated into learning. humanoid.guide covers Robbyant’s tactile foundation model “LingBot-Touch” reported on a tactile foundation model from another company. Viewed alongside reports on Sharpa’s products and tactile applications, hardware, tactile information and learning models are related areas. However, these reports alone do not establish a technical connection between Sharpa’s products and other companies’ models, or whether they are integrated within Sharpa. A shared company name or technology area should be distinguished from confirmed cooperation. To assess this publication as progress in industry implementation, the key questions are whether public information confirms the subject covered in the article, its actual form of availability, compatible robots and sensors, evaluation conditions, results from repeated tests, and operational status in the field. In particular, even when an example of learning using tactile sensing appears alongside a model description, that alone does not substantiate how the training data was obtained or whether the approach can be applied to other robots. These points cannot be determined from the fact of publication and require confirmation through public materials. See the original article for details.