ROHM Semiconductor On-Device Anomaly Detection Starter Kit
ROHM Semiconductors On-Device Anomaly Detection Starter Kit for Edge AI combines an AI-enabled MCU, accelerometer, USB interface, and real-time visualization. ROHM’s Solist-AI™ is an on-device AI solution for Edge AI that enables learning and inference directly on a device without requiring a server, cloud service, or network connection. The SK-AD06-D63Q2557TB Solist-AI Starter Kit provides a ready-to-use platform for evaluating this solution through acceleration-based anomaly detection.
The kit expands the RB-D63Q2557TB64 reference board with an accelerometer board, USB interface, connection cables, a preprogrammed demonstration, and PC-based visualization. Instead of first selecting and connecting the additional hardware required for an anomaly detection demonstration, users can begin evaluating the complete process, from normal-condition learning to real-time anomaly monitoring, using a prepared system configuration. A separate debug adapter is required to rebuild the sample project, modify the firmware, or rewrite the program.
The RB-D63Q2557TB64 is suited to developers who want direct access to the ML63Q2557 Solist-AI MCU and plan to build a customized sensor and interface configuration. The SK-AD06-D63Q2557TB includes the same reference board and adds an acceleration sensor evaluation board, an MM-FT232HC multifunction USB conversion module, jumper cables, and a USB cable. This integrated configuration reduces the hardware preparation required for initial evaluation and gives users a practical starting point for exploring vibration and movement anomaly detection.
During evaluation, users first collect acceleration data under normal conditions and train the system locally using the Solist-AI solution. Users can then move or shake the sensor board to create a different condition and observe the resulting anomaly score. Solist-AI Scope displays acceleration waveforms, learning status, and anomaly scores in real time, helping users understand how changes in movement or vibration are identified without first developing a custom PC visualization tool.
The accelerator supports high-speed AI processing, including operations useful for processing vibration sensor data, while reducing the CPU processing load. The ML63Q2557 supports AI processing at approximately 40mW. This allows engineers to evaluate on-device AI while keeping sensor data and AI processing at the device level.
In addition to the Solist-AI processing capability, the ML63Q2557 integrates embedded sensing, communication, and control functions. These include a CAN FD controller, three-phase motor-control PWM, two 12-bit successive-approximation A/D converter units with 12 input channels, three analog comparators, I2C, SPI, UART, and DMA. The MCU operates from 2.3V to 5.5V at frequencies up to 48MHz and includes 256Kbytes of program Flash, 8Kbytes of data Flash, and 16Kbytes of RAM.
Features
- Ready-to-use Solist-AI Starter Kit for acceleration-based anomaly detection evaluation
- Includes the RB-D63Q2557TB64 reference board with the ML63Q2557 Solist-AI MCU
- Includes an acceleration sensor evaluation board
- Includes the MM-FT232HC multifunction USB conversion module
- Includes jumper cables and a USB 2.0 Type-A-to-Type-C cable
- Preprogrammed Solist-AI anomaly detection sample for initial functional evaluation
- On-device learning and prediction for edge AI without a server, cloud service, or network connection
- Solist-AI Scope visualization of acceleration waveforms, learning status, and anomaly scores
- Approximately 40mW during AI processing
- 32-bit Arm® Cortex®-M0+ processor operating at up to 48MHz
- Integrated CAN FD, motor-control PWM, 12-bit ADCs, analog comparators, I2C, SPI, UART, and DMA
- Separate debug adapter required for firmware modification and program rewriting
Applications
- Predictive maintenance concept evaluation
- Motor and bearing anomaly detection evaluation
- Vibration-based condition monitoring
- Industrial equipment condition monitoring
- Household appliance condition monitoring
- Factory automation sensor evaluation
- Solist-AI and on-device machine-learning prototyping
- Accelerometer-based anomaly detection demonstrations
- Embedded sensing and control proof-of-concept development
Kit Contents
- One Solist-AI starter kit start guide
- One RB-D63Q2557TB64 reference board equipped with Solist-AI MCU (ML63Q2557)
- One KXTJ3-1057-EVK-001 acceleration sensor board
- One MM-FT232HC multi-function USB conversion module
- Four jumper cables (female-to-female), 15cm (red/black/blue/green) (one of each color)
- One USB cable, USB 2.0 Type-A male to USB Type-C® male (0.5m length)
Additional Resources
Connection Diagram
