AM263Px MCU+ SDK  26.00.00
IMU Sensor Data Capture

Table of Contents

Introduction

This example demonstrates real-time accelerometer-based activity classification using the BMI270 sensor on the TIDA-010997 booster pack. It showcases both data capture for dataset building and live inference preview capabilities, enabling seamless integration with Edge AI Studio.

The following modes are supported:

  • Live Capture (DATA_ACQUISITION): Streams raw 3-axis accelerometer readings to build datasets for training custom models
  • Live Preview (SENSOR_INFERENCE): Runs the trained classification model and streams inference results in real-time

Supported Combinations

Parameter Value
CPU + OS r5fss0-0 nortos
Toolchain ti-arm-clang
Board am263px-lp
Example folder examples/ai/imu_sensor_data_capture/

Hardware Requirements

  • TIDA-010997 Edge AI Sensor Boosterpack

External Connections

  • Mount TIDA-010997 Edge AI Sensor Boosterpack on BoosterPack Headers Site 1 (J1/J3 and J2/J4).

Application Overview

This example supports dual-mode operation controlled by Edge AI Studio:

  1. DATA_ACQUISITION mode: Streams raw 3-axis accelerometer samples (6 bytes per sample: X, Y, Z as INT16)
  2. SENSOR_INFERENCE mode: Collects frames of 128 samples, applies feature extraction (FFT, binning, log scaling), runs inference using the trained classification model, and outputs classification results

The inference classifies motion into two classes:

  • Class 0: Jerk
  • Class 1: Smooth

Test vectors are provided for validation and reference testing purposes.

Dependencies

  • Edge AI Studio (for live capture and preview modes)
  • BMI270 sensor hardware (TIDA-010997 booster pack)

Steps to Run the Example

  • When using CCS projects to build, import the CCS project for the required combination and build it using the CCS project menu (see Using SDK with CCS Projects).
  • When using makefiles to build, note the required combination and build using make command (see Using SDK with Makefiles)
  • Connect the TIDA-010997 booster pack with BMI270 sensor
  • Launch a CCS debug session and run the executable, see CCS Launch, Load and Run
  • For live capture or live preview mode, connect Edge AI Studio to the device via UART

See Also

AI Examples

Sample Output

==========================================================
[IMU] IMU Sensor Data Capture Example (Dual-Mode DAP)
==========================================================
[IMU] Feature extraction initialized
[IMU] Variables: 3 (axes)
[IMU] Frame size: 128 samples
[IMU] Model input size: 384
[IMU] Model output size: 2 classes
[IMU] Running test vector validation...
[IMU] Test vector validation PASSED!
[IMU] Model output: [45, -23], Golden: [45, -23]
[IMU] BMI270 sensor initialized (Accel: +/-8g, 100Hz)
[IMU] DAP initialized. Waiting for Edge AI Studio...
[IMU] Supported modes: DATA_ACQUISITION, SENSOR_INFERENCE
==========================================================