Murata links sensing hardware with AI interfaces

Murata Manufacturing has formed a business alliance with Melt Interface Technologies to develop sensing and input hardware for AI and human-machine systems. The first collaboration combines Melt Interface’s ContactGlove3 Pro hand-tracking system with a compact three-axis low-frequency antenna developed by Murata, targeting robot learning, remote operation, and the capture of skilled human movements in manufacturing.

ContactGlove3 Pro uses electromagnetic fields to estimate the position and orientation of a wearer’s hands and fingers. Murata’s three-axis LF antenna forms part of that tracking system, with the company citing stable position and orientation estimation alongside reduced size and power consumption. The integrated glove is due to be demonstrated at CEATEC 2026 in October.

Electromagnetic tracking provides an alternative to purely optical or inertial methods for converting human movement into machine-readable data. A camera system depends on line of sight and image interpretation, while inertial sensors can accumulate positional error over time. Electromagnetic methods measure movement against a generated field, allowing pose information to be captured without requiring every hand or finger to remain visible to a camera.

The trade-offs move into field generation, antenna design, calibration, and the surrounding electromagnetic environment. Nearby conductive or magnetic materials can alter the field, while the antenna has to remain compact enough for a wearable device without sacrificing the signal needed to estimate position and orientation reliably. Mechanical placement and calibration consequently become part of the sensing system rather than packaging details added after the electronics are complete.

A three-axis antenna gives the tracking system information about field components in several directions. That supports pose estimation as the hand changes orientation, while reducing the dimensions and power demand of the antenna helps limit the burden on the wearable hardware. A tracking glove intended to record a skilled manual task becomes less useful if weight, cable routing, battery size, or sensor packaging materially changes the movement being captured.

The manufacturing applications identified by the partners centre on turning human activity into data. Robot learning can use recorded hand motion as an input for training or programming, while remote operation requires movement to be translated into machine control with sufficiently low latency and predictable behaviour. Skill capture raises a related problem: movements performed routinely by experienced operators have to be measured with enough spatial and temporal resolution to separate useful technique from incidental motion.

That places considerable responsibility on the sensing layer before an AI model is introduced. Inconsistent pose estimates, missing samples, calibration drift, or latency become part of the training dataset and can be difficult to distinguish later from limitations in the learning algorithm. Antennas, analogue electronics, sensor geometry, synchronisation, communications, and calibration therefore remain part of an AI system’s performance even when the attention shifts towards software.

Murata also brings experience in component miniaturisation, volume manufacturing, and quality assurance. A wearable tracking demonstrator can tolerate calibration and assembly processes that become impractical at scale. Commercialising the interface would require repeatable antenna characteristics, controlled mechanical tolerances, production test methods, and calibration procedures that can be reproduced without turning every finished unit into a laboratory exercise.

The alliance extends beyond the glove. Murata and Melt Interface are developing a mask voice clip using Murata piezoelectric technology as another input method and plan further work around hand tracking and related interfaces. No production volumes, customer programmes, or qualification timetable have been disclosed, so the collaboration remains an early product-development programme rather than a volume-product announcement.

Murata has recently explored other combinations of sensing and inference, including radar and camera fusion with Smart Eye. The ContactGlove work addresses a different market, but both projects depend on the same boundary between physical sensing and software interpretation: richer algorithms are only useful when the hardware feeding them produces consistent information.

The CEATEC demonstration will show the first integrated result of the new alliance. Beyond that, the engineering challenge is to establish whether the electromagnetic tracking system retains useful accuracy around industrial machinery, tools, metal structures, and repeated operator movement, then translate that behaviour into hardware that can be manufactured and calibrated predictably.