TAIPEI, Nov. 26, 2025 /PRNewswire/ -- With artificial intelligence rapidly advancing toward end devices, Edge AI has become a critical technology driving intelligent and IoT applications. Compared to cloud-based processing, Edge AI enables real-time inference directly on the sensor node or local MCU, effectively reducing latency, enhancing privacy, and lowering power consumption. This makes it particularly suitable for scenarios requiring fast response time, such as gesture control, motion recognition, and equipment monitoring.
Leveraging the high performance, low-power design, and rich peripherals of AT32 MCUs—combined with the Edge Impulse platform—ARTERY Technology continues to accelerate the deployment of multi-sensor Edge AI technology. Developers can integrate various AI functions with shorter development cycles, reducing the time to market and mass production for new products.
To further enhance developers' integrated experience in Edge AI, ARTERY Technology officially launches the AT32 Edge AI Sensor EV Board. Powered by the AT32F403A MCU, the board integrates a variety of sensors including TOF, IMU, magnetometer, ambient light, temperature & humidity, and barometric pressure, and fully supports Edge Impulse model deployment. Developers can run gesture classification, motion recognition, and anomaly detection AI models directly on the device, accelerating product design for AIoT and smart applications.
Three Key Edge AI Applications
1. TOF Gesture Recognition
Based on the onboard VL53L7CX TOF sensor, the AT32 Edge AI Sensor EV Board detects a 4×4 depth array within a 5–20 cm range and uses centroid-weighted algorithms with Edge Impulse neural network models to perform four-direction gesture recognition (up/down/left/right). The OLED can simultaneously display gesture trajectories, making it suitable for smart appliances, HMI interfaces, and in-vehicle controls requiring low-latency, contactless operation.
2. IMU Vibration Anomaly Detection
Using data from the onboard LSM6DS3TR accelerometer and gyroscope, a K-means self-learning model automatically builds "normal operation signatures" to detect abnormal vibrations in fans, motors, or equipment in real time. This is ideal for industrial equipment health monitoring, predictive maintenance, and environmental sensing.
Features:
- 8-D feature extraction (mean, variance, RMS, kurtosis, etc.)
- Local MCU inference without cloud dependence
- Self-learning model that automatically generates anomaly thresholds
3. IMU Motion Classification (Edge Impulse Pipeline)
The AT32 Edge AI Sensor EV Board supports training and deploying IMU-based motion classification models using Edge Impulse. It can recognize movements such as up, down, left, right, circle, and idle, making it suitable for wearables, motion-sensing interaction, and smart control. Models are directly deployable on the AT32F403A and LSM6DS3TR sensor and can be optimized through CMSIS-DSP/NN for improved MCU-based inference performance.
Complete Edge AI Workflow Support
The AT32 Edge AI Sensor EV Board includes a full Edge Impulse development pipeline, covering:
- Sensor data acquisition (TOF / IMU)
- Feature extraction (Spectral Analysis / Feature Engineering)
- Model training for classification and anomaly detection (Neural Networks, K-means)
- MCU-side deployment (EON Compiler / TensorFlow Lite for Microcontrollers)
- Real-time inference output via OLED and UART
Faster AI Development and Mass Production
With the AT32 Edge AI Sensor EV Board, developers can quickly validate multi-sensor AI models at minimal cost and deploy them directly on end devices, accelerating the overall development journey from prototype to mass production. ARTERY will continue to provide more AI models, algorithms, and tool resources to help the industry further popularize edge intelligence and explore new possibilities for AIoT innovation.
Looking Ahead
ARTERY Technology will continue strengthening R&D in Edge AI while enhancing integration between high-performance MCUs and AI algorithms, enabling global customers to build smarter, low-power, and competitive products. As more sensor, voice, and imaging applications emerge, ARTERY will collaborate with partners to advance Edge AI adoption, accelerate smart-industry upgrades, and shape a more efficient and sustainable technological future.
Related Technical Documents (Application Notes):
AN0286 : Introduction to Edge AI Sensor EV Board
AN0287 : IMU K-means Anomaly Detection
** The press release content is from PR Newswire. Bastille Post is not involved in its creation. **
ARTERY Announces AT32 Edge AI Sensor EV Board for Multi-Sensor Gesture, Motion, and Anomaly Detection
