When the AI model runs on the sensor, anomaly detection happens without a cloud round-trip. At IMTS 2026, Tim Gicewicz of TDK SensEI explained how edgeRX collects vibration and temperature data from motors, pumps, and gearboxes, builds a baseline of normal equipment behavior on the device, and flags deviations locally.
Why Run AI on the Sensor Instead of the Cloud?
Edge AI processes data at the source, eliminating cloud round-trips for anomaly detection. The sensor builds a baseline model of normal equipment behavior from vibration amplitude scores and flags deviations locally, delivering alerts through a dashboard and email without depending on a cloud connection.
The edgeRX Pro performs this on-device detection using quantized neural networks, tiny convolutional networks, and autoencoders running on low-power microcontrollers. When connected through USB wired power instead of battery, the sensor supports higher sampling frequencies and continuous execution of more complex AI models.
What Can an Edge AI Sensor Detect?
The edgeRX Pro, launched in August 2026, expands detection beyond vibration and temperature to compressed air leaks, acoustic anomalies, shaft misalignment, bearing defects, and oil and gas leaks. Four sensing modalities work together in a single IP67-rated device with up to 10 years of battery life, mounting on equipment without current clamps or invasive wiring.
| Sensing Modality | What It Detects | Applications |
| 6-axis IMU (accelerometer + gyroscope) | Vibration, rotational motion | Bearing defects, rotor bar breakage |
| Magnetometer | Magnetic field changes | Phase imbalance, eccentric air gap |
| Digital microphone | Acoustic anomalies | Compressed air leaks, mechanical faults |
| Temperature sensor | Thermal deviations | Overheating, friction buildup |
These capabilities apply across manufacturing and energy sectors including lithium battery production, semiconductor fabrication, precision machining, steel processing, electric power generation, natural gas distribution, and HVAC. edgeRX was named 2026 Industrial IoT Product of the Year by IoT Evolution.
How Quickly Does an edgeRX Sensor Deploy?
Sensors deliver immediate readings upon placement on a motor, pump, or gearbox. The AI model needs approximately two weeks of operational data to build the baseline that separates normal equipment behavior from anomalies. Once the baseline is established, the system sends alerts through dashboard and email when conditions deviate from the learned pattern.
This article is based on a video interview with Tim Gicewicz, Sr. Solutions Architect at TDK SensEI, and Lucian Fogoros of IIoT World, recorded at IMTS 2026. AI tools were used to help summarize and organize the content. Reviewed and edited by the IIoT World editorial team.
Sponsored by TDK SensEI.
Sources:
