Question/Issue:
Im currently training accelerometer based classifier to classify motion. I achieve descent accuracy during training > 90%. When i deploy the code via Arduino IDE, The deployment is successful. But the inference gets stuck after IMU initialization.
18:02:58.994 → Edge Impulse Inferencing Demo
18:03:00.222 → IMU initialized
I used the test code that comes with the downloaded library after building the code on Edge Impulse platform. I was wondering why this happens and any ideas to debug this?
Project ID:
vibration_sensor
Context/Use case:
[Provide context or use case where the issue is encountered]
Summary:
I’m currently working on an Edge Impulse project where I’m training an accelerometer-based motion classifier using the Arduino Nano 33 BLE Sense Rev2. The model trains well and achieves over 90% accuracy on the test dataset.
After building the model on the Edge Impulse platform, I downloaded the Arduino library and integrated the nano_ble33_sense_accelerometer_continuous example into the Arduino IDE. The project compiles and uploads successfully to the board.
However, when I run the sketch, it prints:
Edge Impulse Inferencing Demo
IMU initialized
And then it just hangs — no further output, no inferences are run.
Here’s what I’ve tried:
Confirmed that the board is correctly selected in the IDE (Nano 33 BLE Sense Rev2)
Recompiled and reuploaded the code
Verified that the IMU does initialize successfully (via the debug print)
Used the default example sketch provided with the Edge Impulse export
Has anyone experienced similar behavior or have suggestions on how to debug this? I’m wondering if:
There’s a blocking call after the IMU initialization
The loop is waiting on data that’s never arriving
Some initialization step is silently failing after setup
Any ideas or debugging tips would be appreciated!
Steps to Reproduce:
Expected Results:
I expect to see continuous classification and inference running
Actual Results:
The system gets stuck
Reproducibility:
- [ ] Always