Wetour Robotics shows “conductor” neural wristband turning muscle outlines into real-time hand digital twins, citing Meta’s open emg2pose dataset

Wetour Robotics, a company whose shares trade on the Nasdaq under the ticker WETO, demonstrated a neural wristband designed to translate human muscle activity into computer-understandable representations of hand motion and intent. In a presentation carried in a market-news post, the company said its system, called the Conductor Neural Wristband, uses training powered by Meta’s open emg2pose dataset.

According to the announcement description, the wristband demo focuses on converting indicates generated by wrist muscle activity into a real-time, three-dimensional “hand digital twin,” essentially a live virtual model of a user’s hand movement that can be updated moment by moment. The company also described gesture-to-text command functionality as part of the same human-machine interaction pipeline.

The technology pitch centers on what the announcement frames as an “on-device human-intent data layer” for robotics, meaning an approach to capturing what a person is trying to do and converting it into inputs that robots or software can act on. Rather than relying solely on external cameras or microphones, the wristband concept aims to interpret bodily indicates directly at the body, which the demo suggests can support low-latency interaction.

The market-news post attributes the wristband’s training to Meta’s open emg2pose dataset, an open dataset Meta has made available for research into mapping electromyography (EMG) indicates, which reflect muscle electrical activity, to hand poses (the configuration of a hand and fingers). In the wetour demonstration description, this dataset is presented as the foundation that enables the wristband to learn how to associate muscle patterns with hand motion.

For Meta, opening the emg2pose dataset fits within a broader strategy around advancing research and tooling for perception and physical-world AI. Datasets like emg2pose are intended to help accelerate work on reading bodily indicates and converting them into structured representations, which can be reused by developers building applications that go beyond conventional text or image inputs.

In the wetour demonstration description, the company did not provide additional specifics that investors and engineers typically look for, such as the number of users or sessions used for training, the model architecture behind the wristband’s inference, the achieved accuracy for pose estimation, or the latency measured during real-time operation. It also did not disclose hardware details like sensor type, sampling rate, battery life, or whether the processing is fully on-device or partially offloaded.

The announcement similarly did not offer clarity on commercialization milestones, such as whether the Conductor Neural Wristband is available for pilots, how pricing would be structured for robotics customers, or what evaluation targets Wetour is pursuing with partners. For now, the disclosure is framed as a live demonstration rather than a product launch with formal performance benchmarks or contractual terms.