All boards come with an accelerometer. An accelerometer is an electromechanical device that will measure acceleration forces. These forces may be static, like the constant force of gravity pulling at your feet, or they could be dynamic - caused by moving or vibrating the accelerometer.

Acceleration is measured in units of gravities (g) or units of m/s2. One g unit = 9.81 m/s2.

The specific accelerometer model varies amongst the boards, howevever the API provides accelerometer agnostic functions in the accelerometer.h header file that can be safely used with all supported accelerometers.

Users can check which accelerometer is on their board at runtime to determine the appropriate accelerometer specific functions they need to use, if necessary.

accType = libmetawear.mbl_mw_metawearboard_lookup_module(board, MBL_MW_MODULE_ACCELEROMETER)
switch (accType) {

Acceleration Sampling

Acceleration sampling measures the current acceleration forces at periodic intervals. To enable acceleration sampling, call mbl_mw_acc_enable_acceleration_sampling before starting the accelerometer.

Linear acceleration is represented with the MblMwCartesianFloat struct and the values are in units of Gs. The x, y, and z fields contain the acceleration in that direction.

accel_data_signal= libmetawear.mbl_mw_acc_bosch_get_acceleration_data_signal(board)

libmetawear.mbl_mw_datasignal_subscribe(accel_data_signal, None, sensor_data_handler)
libmetawear.mbl_mw_acc_bosch_set_range(board, AccBoschRange._4G)



Users can configure the output data rate and the sampling range; these parameters control the sampling rate and the data range and resolution respectively. After configuring the settings, call mbl_mw_acc_write_acceleration_config to write the configuration to the sensor.

# Set ODR to 25Hz or closest valid frequency
libmetawear.mbl_mw_acc_set_odr(board, 25.0)

# Set range to +/-4g or closest valid range
libmetawear.mbl_mw_acc_set_range(board, 4.0)

# Write the config to the sensor

High Frequency Stream

Firmware v1.2.3+ contains a packed mode for the accelerometer which combines 3 acceleration data samples into 1 ble packet allowing the board to stream data at a throughput higher than 100Hz. This special data signal is retrieved from the mbl_mw_acc_get_high_freq_acceleration_data_signal function and is only for streaming; do not use it with data processing or logging.

accel_data_signal= libmetawear.mbl_mw_acc_get_high_freq_acceleration_data_signal(board)

libmetawear.mbl_mw_datasignal_subscribe(accel_data_signal, None, sensor_data_handler)

libmetawear.mbl_mw_acc_set_odr(board, 200.0)


Wrist Wear Gestures

The BMI270 is designed for Wear OS by GoogleTM7 and features wrist gestures such as flick in/out, push arm down/pivot up, wrist jiggle/shake that help navigate the smartwatch.


For flick-in detection, the user must slowly turn the wrist away from the body (i.e. roll-out shown with a light-grey arrow in) and then quickly bring it back (i.e. roll-in shown with a darker-black arrow in to its original position.

For flick-out detection, the user must quickly turn the wrist away from the body (i.e. roll-out shown with a darker-black arrow in above picture) and then slowly bring it back (i.e. roll-in shown with a light-grey arrow in above picture) to its original position.

The speed of the roll-out and roll-in movements determine if the user performed a flick-in or a flick-out movement.

// Start the accelerometer
// Configure
libmetawear.mbl_mw_acc_bmi270_wrist_gesture_armside(board, 0) // left arm
// Get gesture signal
gesture_signal = libmetawear.mbl_mw_acc_bmi270_get_wrist_detector_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(gesture_sig, None, sensor_data_handler)
// Start detecting motion and turn on acc

There are config functions for the wrist wear feature:


Activity Detector

The BMI270 can detect simple user activities (unknown, still, walking, running) and can send an interrupt if those are changed, e.g. from walking to running or vice versus.

// Start the accelerometer
// Get activity signal
activity_sig = libmetawear.mbl_mw_acc_bmi270_get_activity_detector_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(activity_sig, None, sensor_data_handler)
// Start detecting motion and turn on acc

Wrist Wear Wakeup

The BMI270 has a wrist wear wakeup feature that is designed to detect any natural way of user moving the hand to see the watch dial when wearing a classical wrist watch.

The feature is intended to be used as wakeup gesture (i.e. for triggering screen-on or screen-off) in wrist wearable devices.

This feature has dependency on the device orientation in the user system. Implementation of the feature to detect gesture assumes that the sensor co-ordinate frame is aligned with the device/system co- ordinate frame. The assumed default device/system co-ordinate frame is depicted below.

Please refer to this section regarding axis remapping.

// Start the accelerometer
// Get gesture signal
wrist_sig = libmetawear.mbl_mw_acc_bmi270_get_wrist_detector_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(wrist_sig, None, sensor_data_handler)
// Start detecting motion and turn on acc

There are config functions for the wrist wear feature:


Motion Detector

The BMI270 can detect significant motion (android motion), any motion (high acc motion) or no motion. The accelerometer must be at least running at 25Hz.

Detect Any Motion

The anymotion detection uses the slope between two acceleration signals to detect changes in motion.

// Start the accelerometer
// Set any motion config - acc must be on for this
libmetawear.mbl_mw_acc_bosch_set_any_motion_count(board, 5)
libmetawear.mbl_mw_acc_bosch_set_any_motion_threshold(board, 170.0)
libmetawear.mbl_mw_acc_bosch_write_motion_config(board, MBL_MW_ACC_BOSCH_MOTION_ANYMOTION)
// Get any motion signal
any_motion = libmetawear.mbl_mw_acc_bosch_get_motion_data_signal(board);
libmetawear.mbl_mw_datasignal_subscribe(any_motion, None, sensor_data_handler)
// Start detecting motion
libmetawear.mbl_mw_acc_bosch_enable_motion_detection(board, MBL_MW_ACC_BOSCH_MOTION_ANYMOTION)

Detect No Motion

The nomotion detection can detect when there is no motion for a certain amount of time.

libmetawear.mbl_mw_acc_bosch_set_no_motion_count(board, 5)
libmetawear.mbl_mw_acc_bosch_set_no_motion_threshold(board, 144.0)
libmetawear.mbl_mw_acc_bosch_write_motion_config(board, MBL_MW_ACC_BOSCH_MOTION_NOMOTION)
no_motion = libmetawear.mbl_mw_acc_bosch_get_motion_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(no_motion, None, sensor_data_handler)
libmetawear.mbl_mw_acc_bosch_enable_motion_detection(board, MBL_MW_ACC_BOSCH_MOTION_NOMOTION)

Detect Significant Motion

The significant motion interrupt implements the interrupt required for motion detection in Android 4.3 and greater: A significant motion is a motion due to a change in the user location.

Examples of such significant motions are walking or biking, sitting in a moving car, coach or train, etc. Examples of situations that does typically not trigger significant motion include phone in pocket and person is stationary or phone is at rest on a table which is in normal office use. .

libmetawear.mbl_mw_acc_bosch_set_sig_motion_blocksize(board, 250)
libmetawear.mbl_mw_acc_bosch_write_motion_config(board, MBL_MW_ACC_BOSCH_MOTION_SIGMOTION)
sig_motion = libmetawear.mbl_mw_acc_bosch_get_motion_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(sig_motion, None, sensor_data_handler)
libmetawear.mbl_mw_acc_bosch_enable_motion_detection(board, MBL_MW_ACC_BOSCH_MOTION_SIGMOTION)

Step Counter

The BMI160 accelerometer comes with a built in step counter. It has three operation modes that configure the sensitivity and robustness of the counter:




Balanced between false positives and false negatives, recommended for most applications


Few false negatives but eventually more false positives, recommended for light weighted people


Few false positives but eventually more false negatives

When you have set the operation mode, call mbl_mw_acc_bmi160_write_step_counter_config to save the configuration to the board.

libmetawear.mbl_mw_acc_bmi160_set_step_counter_mode(board, AccBmi160StepCounterMode.NORMAL)

The BMI270 accelerometer does not support step counter modes.

Reading The Counter

One way to retrieve step counts is to periodcally read the step counter. To read the step counter, call mbl_mw_datasignal_read with the step counter data signal.

The counter is not enabled by default so you will need enable it by calling mbl_mw_acc_bmi160_enable_step_counter when configuring the board.

step_counter_signal = libmetawear.mbl_mw_acc_bmi160_get_step_counter_data_signal(board)

libmetawear.mbl_mw_datasignal_subscribe(step_counter_signal, None, sensor_data_handler)

For the BMI270, you can call mbl_mw_acc_bmi270_enable_step_counter when configuring the board.

Using The Detector

Alternatively, you can receive notifications for each step detected by calling mbl_mw_acc_bmi160_enable_step_detector instead.

step_detector_signal= libmetawear.mbl_mw_acc_bmi160_get_step_detector_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(step_detector_signal, None, sensor_data_handler)

For the BMI270, the detector will not send notifications every step but instead every 20*X steps:

libmetawear.mbl_mw_acc_bmi270_set_step_counter_trigger(board, 1) //every 20 steps
detector = libmetawear.mbl_mw_acc_bmi270_get_step_detector_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(detector, None, sensor_data_handler)

Orientation Detection

The orientation detector alerts you when the sensor’s orientation changes between portrait/landscape and front/back. Data is represented as an MblMwSensorOrientation enum.

This feature is currently only supported on devices using the BMI160 or BMA255 accelerometers. It is not supported on the BMI270.

orientation = libmetawear.mbl_mw_acc_bosch_get_orientation_detection_data_signal(board)
libmetawear.mbl_mw_datasignal_subscribe(orientation, None, sensor_data_handler)