Sensors Converge 2026: Smarter and lower-power sensors

The Sensors Converge 2026 conference showcased some of the latest advances in sensor and sensing solutions for applications ranging from wearables and smartphones to industrial and automotive. The show, with over 160 exhibitors, also highlighted the industry’s shifting focus to edge AI and smart, connected systems with demos that showcased real-world applications in edge AI, robotics, and autonomous systems.

While sensor manufacturers continue to focus on shrinking solutions and package sizes, this year’s product introductions also indicate an increased need for lower power consumption. Here is a sampling of new sensors featured at this year’s show.

Vibration sensors across wearables and industrial

Upbeat Technology showcased its latest family of low-power MEMS vibration sensors and vibration processing units (VPUs), including the UPM01 and UPM02 series with a UP201/301 dual-core RISC-V AI microcontroller (MCU), aimed at high-quality voice clarity and predictive intelligence in a small footprint.

Suited for space-constrained wearables applications, the UPM01/UPM02 VPU, also called a bone-conduction microphone, measures 3.2 × 2.5 mm, and the UP201 dual-core RISC-V AI MCU measures 3.0 × 3.0 mm. Together, they create Upbeat’s Tiny AI Engine that provides on-device intelligence to wearables, industrial systems, drones, and consumer electronics. The solution enables “crystal-clear voice” in open wearable stereo (OWS) headsets, smart glasses, and intelligent voice recorders and delivers predictive maintenance for industrial automation.

The UPM01 series offers multiple interface variants: the UPM01A (analog), UPM01Ax (higher-sensitivity analog), UPM01D (digital), and UPM01Dx (higher-sensitivity digital). The UPM02 provides analog and digital options with a higher signal-to-noise ratio (SNR) for applications in which audio clarity is critical, the company said.

The UPM01 extends the frequency response of conventional MEMS vibration sensors from 5 Hz to 11.3 kHz and delivers an SNR of 60 dB(A) for a more accurate sound capture, while the UPM02 offers a frequency response range from 5 Hz to 5.4 kHz and an exceptionally high SNR of up to 68 dB(A).

Both series consume minimal power and can operate for extended periods on a single battery charge, making them suited for mobile devices, wearables, and other battery-powered applications.

The UP201/UP301 heterogeneous dual-core RISC-V edge AI platform targets energy-efficient deep-learning applications, enabling AI analysis closer to the data source for fast response and lower bandwidth usage. Delivering ultra-low-power, always-on intelligence, the platform enables continuous sensing with minimal power and instant wake-up for intensive AI tasks.

Mass-production shipments for the UPM01/UPM02 have started, with the UP201/UP301 scheduled to ship in October 2026.

Upbeat also unveiled its UP301 + UPM01 Falcon Demo Kit, described as a ready-to-run evaluation platform for machine-vibration analysis. Aimed at engineers who want to prototype and validate predictive maintenance solutions, the kit includes a UP201 dual-core RISC-V AI MCU EVB, variable-speed motor, two UPM01D FPCs, power adapter, and access to the Falcon graphical user interface (GUI), the Upbeat Vibration Analysis Suite GUI software. The demo kit is available for purchase at www.upbeattechtw.com/products/demo-kits.

Other demonstrations included OWS headsets, smart glasses with AI voice interaction, a smart AI voice recorder, a factory machine-vibration application, and smart AI toys with touch-gesture recognition.

Ahead of the show, STMicroelectronics announced its wide-bandwidth, three-axis vibration sensor, aimed at saving space and energy in industrial and automotive condition-monitoring applications. With an extended temperature range of −40°C to 125°C, the IIS3DWBG1 enables vibration monitoring in harsh environments.

The IIS3DWBG1 offers a selectable, full-scale acceleration range of ±2/±4/±8/±16 g and can measure accelerations with a bandwidth up to 6 kHz with an output data rate of 26.7 kHz. Housed in a 2.5 × 3-mm LGA-14L package, the MEMS sensor is suitable for industrial condition-monitoring systems, in which sensor placement and mounting are critical to measurement accuracy.

The small size and wide operating temperature range allow the flexibility to place small, externally attached sensors at optimal diagnostic locations while enabling integration inside smart motors and smart gearboxes, ST said.

In addition, the low power consumption delivers long-lasting operation in battery-powered applications. The sensor’s wide bandwidth and high resolution simplify capturing patterns associated with defects or wear, as well as equipment setup issues such as looseness and misalignment.

The IIS3DWBG1 can also detect electromechanical vibrations in coils, transformers, snubber capacitors, busbars, connectors, and general vibrations originating in the power electronics module, such as traction inverters. This enables automotive OEMs to extend remote diagnostics to cover power modules, as well as traction inverters in electric vehicles.

Thanks to a flat frequency response from DC to above 6 kHz (−3 dB point) and noise density of 75 µg/√Hz in three-axis mode, the sensor detects extremely small vibrations, providing enhanced early warning to prevent equipment failures. The sensor is highly resistant to mechanical shocks, according to ST, and integrates digital features including a configurable low-pass or high-pass filter with selectable cutoff frequency, an embedded FIFO, interrupts, a temperature sensor, and self-test capability.

The IIS3DWBG1 is in production now. An evaluation kit is available.

AMR and TMR sensors

Murata Manufacturing Co. Ltd. introduced its ultra-low-power anisotropic magnetoresistance (AMR) sensors, the MRMS166R and MRMS168R. These sensors are designed to increase battery life in healthcare, wearable, and IoT devices. The MRMS166R is claimed as the first AMR sensor to combine an average current consumption of 20 nA with operation from a 1.2-V supply, enabling extended battery life in coin-cell-powered systems.

These solid-state magnetic sensors detect the presence or absence of a magnetic field and generate an output signal that system logic uses to control functions such as transitions between active and sleep modes. This provides contactless switching without mechanical components, improved reliability, and support for sealed, miniaturized designs, Murata said.

This automatic switching between active and sleep modes is widely used in battery-powered devices to reduce standby power consumption and extend operating life, Murata said. Applications include healthcare, such as capsule endoscopes and medical patches; wearable devices, including AR glasses and wireless earbuds; and security-related IoT devices, such as door-open/close-detection systems and smart locks.

These devices commonly use silver oxide coin batteries (typically 1.55 V) that place constraints on available capacity and operating voltage. This means AMR sensors used as magnetic switches must minimize current consumption while maintaining stable operation at a low voltage, Murata said.

To address these challenges, Murata redesigned the AMR sensor’s internal circuitry, enabling ultra-low current consumption and operation down to 1.2 V. This significantly reduces battery consumption during standby operation, supporting device operation for more than two years in typical use.

The MRMS166R operates over a 1.2-V to 3.6-V supply range (1.5 V typ.) with an average current consumption of 20 nA and a maximum current output of 1 mA. The MRMS168R operates over a 2.0-V to 3.6-V supply range (3.0 V typ.), with an average current consumption of 80 nA and a maximum output current of 12 mA, providing higher output drive capability for devices requiring increased load current. Both devices are housed in a compact package measuring 1.0 × 1.0 × 0.4 mm (0.04 × 0.04 × 0.02 inches). The MRMS166R and MRMS168R sensors are now in mass production.

MultiDimension Technology Co. Ltd. debuted its tunneling magnetoresistance (TMR) TMR2531 (±1,000-Gauss linear range) and TMR2539 linear sensors (extended ±1,500-Gauss linear range) for smartphone cameras at Sensors Converge. Available in production quantities, these ultra-compact TMR linear sensors are designed for high-precision smartphone optical image stabilization (OIS) applications.

These sensors enable micron-level displacement measurement in voice coil motor (VCM) modules, allowing VCM driver ICs to precisely correct camera shake in real time during photo and video capture, MDT said. They measure the z-axis perpendicular magnetic field amplitude via a Wheatstone full-bridge configuration with four high-SNR TMR elements.

Periscope-style telephoto lenses have pushed OIS precision requirements into the micron scale to control prism positioning over extended motion ranges, MDT said. The new TMR sensor technology addresses these challenges with a high SNR, broad linear measurement ranges, and high immunity to magnetic interference, making it suited for advanced camera autofocus and OIS solutions in flagship smartphones.

Both series offer a 1.0-V to 5.5-V supply voltage and a shielding capability of ±3,000 Gauss for stable operation in interference-prone VCM environments. They are housed in a small DFN4L package (0.8 × 0.5 × 0.25 mm) for constrained VCM designs.

Faster sensor development

TDK Corp. introduced two development tools at Sensors Converge to simplify evaluation of TDK sensors. The InvenSense SensorStage software is an evaluation platform to simplify development and accelerate data analytics for TDK’s SmartMotion inertial measurement units (IMUs) and TMR magnetometers, while SensorGPT uses AI to generate simulated datasets to improve and accelerate development of edge AI IoT devices.

The all-in-one platform SensorStage bridges the gap between simple GUIs and custom test benches, offering advanced visual analytics and automated scripting to help engineers move from setup to insight without manual configuration, TDK said. SensorStage enables evaluation of complex, on-chip algorithms for applications in OIS, wearables, AR/smart glasses, and IoT with a future-proof architecture that supports existing and upcoming high-performance sensors.

The SensorStage platform is paired with the SmartMotion development board. Together, sophisticated on-chip features including machine-learning algorithms, the APEX engine for Gyro Assisted Fusion, motion and event detection, and chip-level power consumption are visualized. This delivers precise calibration and faster time to market for complex designs.

SensorStage is currently available for InvenSense ICM-456xx and ICM-426xx SmartMotion IMUs and will soon be available for additional InvenSense MEMS sensor solutions.

SensorGPT uses generative AI, signal processing, statistical methods, and simulations to create and manage sensor data at scale. Particularly aimed at smart IoT and ambient IoT applications, the AI tool streamlines model development and deployment, reducing time and cost, while enhancing the performance and efficiency of edge AI models and applications, TDK said.

SensorGPT sensor data synthesis trains generative models with limited real-world data to learn underlying patterns and generates synthetic data that mimics real-world data. It reduces the reliance on real-world data through intelligent sensor data synthesis, cutting data-collection efforts from 80% to nearly 10%, according to TDK, which enables faster, more scalable edge AI development.

The AI tool leverages physics-based and mathematical models to simulate and generate synthetic sensor data and uses mathematical and computational techniques to simulate data reflecting the dynamics and characteristics of real sensor outputs, TDK explained.

Other features include data-augmentation techniques that automatically transform existing sensor data into diverse datasets spanning a range of conditions and scenarios, while the assisted annotation streamlines the labeling of training data, which improves the quality for model training.

SensorGPT generates a 90% similarity between synthetic and real-world sensor data. This enables the use of the synthetically generated data for faster edge AI solution prototyping, testing, and deployment. It reduces edge AI model-building time from five-plus months down to a few weeks, according to TDK.

The post Sensors Converge 2026: Smarter and lower-power sensors appeared first on EDN.