OYMotion sdk for Python
OYMotion SDK is the software development kit for developers to access OYMotion products.
See the contributing guide to learn how to contribute to the repository and the development workflow.
MIT
pip install sensor-sdk When a USB Bluetooth HCI dongle is plugged in and usable by libusb (e.g. Actions 10d7:b012 or 33fa:0012; several dongles can be used at once, each serving one connection), the SDK automatically drives it directly with a bumble host-mode backend instead of the OS Bluetooth stack — no Bluetooth permission prompt, works even when the internal Bluetooth is off, and allows larger ATT MTU. The bleak-bumble backend is vendored into the SDK (sensor/bleak_bumble/, MIT license), and the bumble stack + libusb are declared as regular dependencies — a plain pip install sensor-sdk is all that is needed.
Behavior and controls:
- With the dependencies installed and a usable dongle present, the backend is selected automatically on all platforms when the SDK starts (native
bleakis used otherwise). "Usable" means libusb can open the device: on Windows the dongle must be bound to the WinUSB driver, on Linux the user needs udev permission, on macOS USB access is driverless. Check the active backend withSensorController.getBLEBackendName()("bumble"/"bleak"). SensorController.checkSetupDongle()checks dongle readiness and, when no dongle is usable, runs the bundled setup script with elevation — UAC prompt on Windows (binds the WinUSB driver),sudoon Linux (installs udev rules; the password goes through the controlling terminal, or when there is none — e.g. a GUI app launched without a terminal — a terminal-emulator window is spawned to run the script, falling back topkexec; a replug may still be required afterwards) — then re-checks and returns"OK: N"(N = number of usable dongles detected) on success or an"Error: ..."string with the system error detail on failure. macOS needs no setup and is check-only. The call blocks while waiting for the elevation prompt. The same logic also lives insensor_utils.checkSetupDongle(), exported at package top level assensor.checkSetupDongle()for use without a controller instance. The same scripts (setup_dongle_winusb.ps1,setup_dongle_udev.shand their driver/rules files) ship inside the wheel undersensor/tools/and can also be run manually.SENSOR_SDK_BLE_BACKEND=bleakforces the native backend;SENSOR_SDK_BLE_BACKEND=bumbleforces the dongle backend on any platform;SENSOR_SDK_BUMBLE_TRANSPORToverrides the bumble transport spec (e.g.usb:0).- With the bumble backend, each dongle serves one connection at a time: scanning uses a free dongle and is skipped while all dongles are connected; connections fail fast when no dongle is free. Connection timeout is raised to 25s automatically.
Application will obtain bluetooth permission by itself.
from sensor import *SensorControllerInstance = SensorController()
# register scan listener
if not SensorControllerInstance.hasDeviceFoundCallback:
def on_device_callback(deviceList: List[BLEDevice]):
# return all devices doesn't connected
pass
SensorControllerInstance.onDeviceFoundCallback = on_device_callbackUse def startScan(period_in_ms: int) -> bool to start scan
success = SensorControllerInstance.startScan(6000)returns true if start scan success, periodInMS means onDeviceCallback will be called every periodInMS
Use def scan(period_in_ms: int) -> list[BLEDevice] to scan once time
bleDevices = SensorControllerInstance.scan(6000)The asynchronous variant is asyncScan(period_in_ms: int) -> list[BLEDevice].
Use def stopScan() -> None to stop scan
SensorControllerInstance.stopScan()Use property isScanning: bool to check scanning status
isScanning = SensorControllerInstance.isScanningUse property isEnable: bool to check if bluetooth is enable
isEnable = SensorControllerInstance.isEnableUse onEnableCallback to be notified when the bluetooth enable state changes:
SensorControllerInstance.onEnableCallback = lambda enabled: print("bluetooth on" if enabled else "bluetooth off")Use def requireSensor(device: BLEDevice) -> SensorProfile | None to create SensorProfile.
If bleDevice is invalid, result is None.
sensorProfile = SensorControllerInstance.requireSensor(bleDevice)Use def getSensor(deviceMac: str) -> SensorProfile | None to get SensorProfile.
If SensorProfile didn't created, result is None.
sensorProfile = SensorControllerInstance.getSensor(bleDevice.Address)Use def getConnectedSensors() -> list[SensorProfile] to get connected SensorProfiles.
sensorProfiles = SensorControllerInstance.getConnectedSensors()Use def getConnectedDevices() -> list[BLEDevice] to get connected BLE Devices.
bleDevices = SensorControllerInstance.getConnectedDevices()Use def terminate() to terminate sdk
def terminate():
SensorControllerInstance.terminate()
exit()
def main():
signal.signal(signal.SIGINT, lambda signal, frame: terminate())
time.sleep(30)
SensorControllerInstance.terminate()
Please MAKE SURE to call terminate when exit main() or press Ctrl+CPlease register callbacks for SensorProfile
from typing import List
from sensor import SensorProfile, SensorData, DeviceStateEx
sensorProfile = SensorControllerInstance.requireSensor(bleDevice)
# register callbacks
def on_state_changed(sensor: SensorProfile, newState: DeviceStateEx):
# device state transitions (Connecting/Connected/Ready/Disconnected/...)
# called synchronously on the SDK state machine; return quickly
# please do logic when device disconnected unexpected
pass
def on_error_callback(sensor: SensorProfile, reason: str):
# called when error occurs (dongle unplugged, reconnect budget exhausted, ...)
pass
def on_power_changed(sensor: SensorProfile, power: int):
# callback for get battery level of device, power from 0 - 100
# (invalid -1 readings are never reported here; getBatteryLevel() may
# still return -1 when no valid reading is available yet)
# the reported value is stabilized with a hysteresis band (±2%): it only
# changes when a new reading differs from the current value by 2 or more,
# so ±1 jitter is filtered while real drain/charge trends are still tracked
pass
def on_data_callback(sensor: SensorProfile, data_list: List[SensorData]):
# called after start data transfer; each invocation delivers the whole
# batch of SensorData objects parsed together (loop over it to process
# each one)
for data in data_list:
pass
sensorProfile.onStateChanged = on_state_changed
sensorProfile.onErrorCallback = on_error_callback
sensorProfile.onPowerChanged = on_power_changed
sensorProfile.onDataCallback = on_data_callbackCallback threading model:
onDataCallbackruns on a dedicated single-worker thread pool per profile, so batches are delivered strictly in arrival order.onPowerChangedandonErrorCallbackshare a multi-worker thread pool.onStateChangedis invoked synchronously.- Keep callbacks short or thread-safe; update UI through your framework's main-thread mechanism (e.g. Qt signals).
A fifth callback, onAutoReconnect, customizes stream recovery after an abnormal disconnect — see section 15.1.
Use def connect() -> bool to connect.
success = sensorProfile.connect()Use def disconnect() -> bool to disconnect.
If data notification is currently active, disconnect() will automatically stop it first before closing the BLE connection.
success = sensorProfile.disconnect()Use property deviceState: DeviceStateEx to get device status.
Please send command in 'Ready' state, should be after connect() return True.
deviceStateEx = sensorProfile.deviceState
# deviceStateEx has define:
# class DeviceStateEx(Enum):
# Disconnected = 0
# Connecting = 1
# Connected = 2
# Ready = 3
# Disconnecting = 4
# Invalid = 5Use property BLEDevice: BLEDevice to get BLE device of SensorProfile.
bleDevice = sensorProfile.BLEDeviceUse property autoReconnect: bool (default True) to control automatic recovery. While True and the device is streaming, an abnormal disconnect — remote link loss or a long no-data (half-dead link) disconnect — is followed by automatic reconnect → init() with the previous init arguments → re-applying the setParam parameters from the previous streaming session (in the order they were set) → startDataNotification(). Recovery retries on the next successful reconnect if a step fails. Explicit user calls (connect(), disconnect(), stopDataNotification()) cancel the pending resume, and setting autoReconnect = False disables the behavior entirely.
sensorProfile.autoReconnect = True # default; set False to opt outCustom recovery via onAutoReconnect (default None): when the auto-reconnect finds the disconnected device again (back in Ready, about to resume), this callback is invoked instead of the default flow.
def on_reconnect(sensor, restore: bool) -> bool:
# restore=True -> a previous session exists (init args + setParam values can be
# preserved and restored); restore=False -> fresh-init case
# return True -> the app handled recovery itself (fresh init or custom flow);
# the SDK skips the default recovery
# return False -> fall back to the default flow (connect -> init -> replay
# setParam values -> startDataNotification)
sensor.init(32, 60000)
sensor.startDataNotification()
return True
sensorProfile.onAutoReconnect = on_reconnectCallback exceptions are logged and treated as False (fall back). The callback runs on the SDK recovery thread, so blocking calls (init(), setParam(), startDataNotification()) are allowed inside it.
Use def getDeviceInfo() -> DeviceInfo | None to get device info of SensorProfile.
Please call after device in 'Ready' state and init() has succeeded, returns None otherwise.
deviceInfo = sensorProfile.getDeviceInfo()
# deviceInfo is a DeviceInfo object with attributes:
# DeviceName, ModelName, HardwareVersion, FirmwareVersion, MTUSize
# plus a ChannelCount / SampleRate attribute pair for each modality:
# Ppg, Spo2, Impe, Emg, Eeg, Ecg, Acc, Gyro, Brth, MagAngle, Euler, Quat
# e.g. deviceInfo.EmgChannelCount, deviceInfo.EegSampleRateUse def init(packageSampleCount: int, powerRefreshInterval: int) -> bool.
Please call after device in 'Ready' state, return True if init succeed.
success = sensorProfile.init(5, 60*1000)packageSampleCount: set sample counts of SensorData.channelSamples per batch in onDataCallback() powerRefreshInterval: callback period for onPowerChanged()
Use property hasInited: bool to check if init data transfer succeed.
hasInited = sensorProfile.hasInitedUse def startDataNotification() -> bool to start data notification.
Please call if hasInited return True
success = sensorProfile.startDataNotification()Data type list:
class DataType(Enum):
NTF_ACC = 0x1 # acceleration, unit is g
NTF_GYRO = 0x2 # gyroscope, unit is degree/s
NTF_EULER_DATA = 0x4 # euler angle, unit is degree
NTF_QUATERNION = 0x5 # quaternion (w, x, y, z)
NTF_GEST = 0x07 # gesture id
NTF_EMG = 0x8 # unit is uV
NTF_MAG_ANGLE_DATA = 0x0D
NTF_EEG = 0x10 # unit is uV
NTF_ECG = 0x11 # unit is uV
NTF_IMPEDANCE = 0x12 # electrode impedance
NTF_IMU = 0x13 # aggregated IMU batch (acc 0-2 / gyro 3-5 / euler 6-8 / quat 9-12;
# new-EMG devices only, see DeviceInfo.ImuChannelCount)
NTF_ADS = 0x14
NTF_BRTH = 0x15 # respiration, unit is uV
NTF_IMPEDANCE_EXT = 0x16
NTF_SPO2 = 0x17 # SpO2 percentage
NTF_PPG = 0x18 # PPG raw samplesProcess data in onDataCallback. Each invocation delivers a list of SensorData batches parsed together; loop over the list to process each batch.
def on_data_callback(sensor: SensorProfile, data_list: List[SensorData]):
for data in data_list:
if data.dataType == DataType.NTF_EEG:
pass
elif data.dataType == DataType.NTF_ECG:
pass
# process data as you wish
for oneChannelSamples in data.channelSamples:
for sample in oneChannelSamples:
if sample.isLost:
# do some logic
pass
else:
# draw with sample.data & sample.channelIndex
# print(f"{sample.channelIndex} | {sample.sampleIndex} | {sample.data} | {sample.impedance}")
pass
sensorProfile.onDataCallback = on_data_callbackUse def stopDataNotification() -> bool to stop data transfer.
success = sensorProfile.stopDataNotification()Use property isDataTransfering: bool to check if it's data transfering.
isDataTransfering = sensorProfile.isDataTransferingUse def getBatteryLevel() -> int to get battery level. Please call after device in 'Ready' state.
batteryPower = sensorProfile.getBatteryLevel()
# batteryPower is battery level returned, value ranges from 0 to 100, 0 means out of battery, while 100 means full;
# -1 means no valid reading is available yet (onPowerChanged never reports -1).Please check console.py in examples directory
all methods start with async is async methods, they has same params and return result as sync methods.
Please check async_console.py in examples directory
Use def setParam(self, key: str, value: str) -> str to set parameter of sensor profile. Please call after device in 'Ready' state.
The asynchronous variant is asyncSetParam(self, key: str, value: str) -> str.
If the device is already streaming when you change an NTF_* or FILTER_* key, the SDK will stop and restart the data notification so the new setting takes effect immediately.
Below is available key and value:
# Data stream toggles
result = sensorProfile.setParam("NTF_GEST", "ON")
result = sensorProfile.setParam("NTF_EMG", "ON")
result = sensorProfile.setParam("NTF_EEG", "ON")
result = sensorProfile.setParam("NTF_ECG", "ON")
result = sensorProfile.setParam("NTF_IMU", "ON")
result = sensorProfile.setParam("NTF_BRTH", "ON")
result = sensorProfile.setParam("NTF_IMPEDANCE", "ON")
result = sensorProfile.setParam("NTF_MAG_ANGLE", "ON")
result = sensorProfile.setParam("NTF_PPG", "ON")
result = sensorProfile.setParam("NTF_PPG_RAW", "ON") # alias of NTF_PPG
result = sensorProfile.setParam("NTF_SPO2", "ON")
result = sensorProfile.setParam("NTF_GFORCE_EULER", "ON")
result = sensorProfile.setParam("NTF_GFORCE_QUAT", "ON")
result = sensorProfile.setParam("NTF_GFORCE_ACC", "ON")
result = sensorProfile.setParam("NTF_GFORCE_GYRO", "ON")
# set data stream to ON or OFF, result is "OK" if succeed
# NTF_IMU is the master switch of the four NTF_GFORCE_* streams: toggling it
# updates all four, and toggling any of the four updates the aggregated NTF_IMU state.
# Note: on legacy (non-new) EMG devices, NTF_GEST and NTF_EMG are mutually exclusive.
# Firmware filter toggles
result = sensorProfile.setParam("FILTER_50HZ", "ON")
# set 50Hz notch filter to ON or OFF, result is "OK" if succeed
result = sensorProfile.setParam("FILTER_60HZ", "ON")
# set 60Hz notch filter to ON or OFF, result is "OK" if succeed
result = sensorProfile.setParam("FILTER_HPF", "ON")
# set 0.5Hz hpf filter to ON or OFF, result is "OK" if succeed
result = sensorProfile.setParam("FILTER_LPF", "ON")
# set 80Hz lpf filter to ON or OFF, result is "OK" if succeed
# NeuCir remote control (NeuCir devices only)
result = sensorProfile.setParam("NEUCIR_SET_MODE", "APP_REMOTE")
result = sensorProfile.setParam("NEUCIR_APP_CONTROL", "OPEN") # OPEN / CLOSE / STOP
result = sensorProfile.setParam("DEBUG_BLE_DATA_PATH", "d:/temp/test.bin")
# set the bin export path: the session's raw BLE capture is recorded in the system
# temp directory and copied to this location on stopDataNotification / disconnect;
# "True" exports to {DeviceName}_data_YYYYMMDD_HHMMSS.bin in the SDK log directory
# (see setLogPath; disabled when file output is off), "False" or "" disables
# export (the temp bin is just deleted).
# please give an absolute path and make sure it is valid and writeable by yourself
result = sensorProfile.setParam("DEBUG_LOG_PATH", "True")
# enable this profile's log file: {DeviceName}_log_YYYYMMDD_HHMMSS.txt in the SDK
# log directory (see setLogPath), or pass an absolute custom path instead of "True";
# "False" or "" disables it. The profile log contains only this profile's logs plus
# the bleak/bumble logs related to its current connection; all other (common) logs
# go to the controller log.
# getParam("DEBUG_LOG_PATH") returns the current log file path ("" when disabled).Use def getParam(self, key: str) -> str to query the current parameter state of a sensor profile. Please call after the device reaches the 'Ready' state.
The asynchronous variant is asyncGetParam(self, key: str) -> str.
Supported aggregate query keys:
result = sensorProfile.getParam("FILTER")
# Returns a pipe-separated string of all filter states, e.g.:
# "FILTER_50HZ|ON|FILTER_60HZ|ON|FILTER_HPF|ON|FILTER_LPF|ON"
result = sensorProfile.getParam("NTF")
# Returns a pipe-separated string of all notification states, e.g.:
# "NTF_BRTH|ON|NTF_ECG|ON|NTF_EEG|ON|NTF_EMG|ON|..."If the key is not supported, the result starts with "Error".
On every successful connect, the SDK records all raw BLE packets of the session into a .bin file in the SDK log directory (~/Documents/sensorsdklog by default, or the directory of the configured log file), named {DeviceName}_{MAC}_{YYYYMMDD_HHMMSS}.bin. Recording is always on; it is skipped or stopped with a warning when free disk space is below 100MB or a disk error occurs, without affecting live streaming. Bin files can be replayed offline for debugging and packet-loss analysis.
Use def getBinFileInfo(self, file_path: str) -> Optional[dict] to read the config record of a bin file:
info = SensorControllerInstance.getBinFileInfo("path/to/session.bin")
# Returns a dict:
# device_mac, device_name, chip_type, is_universal_stream, feature_map,
# device_info, sensor_datas (per data-type parse config),
# replay_duration (recording seconds, written into the header record at close;
# older bins without a header fall back to a full-file estimate)
# Returns None if the file does not exist or has no config record.Use def replayBinFile(self, file_path: str, sensor: Optional[SensorProfile] = None, realtime: bool = True, timeout: Optional[float] = None) -> Optional[SensorProfile] to replay a bin file through the normal parsing pipeline. Parsed results arrive via onDataCallback on the returned SensorProfile, same as live data.
# Simplest form: controller creates the profile from the bin config record
sensor = SensorControllerInstance.replayBinFile("path/to/session.bin")
# Recommended: reuse an existing profile with callbacks registered
sensor = SensorControllerInstance.requireSensor(device)
sensor.onDataCallback = on_data
SensorControllerInstance.replayBinFile("path/to/session.bin", sensor, realtime=True)sensor: an existing SensorProfile to replay through. WhenNone, the controller creates (or reuses) a profile from the bin config record; in that mode the profile is returned only after replay finishes, so register callbacks on an existing profile to receive data.realtime:Truereplays at the recorded pace;Falsereplays as fast as possible.timeout: seconds to wait for completion.Noneauto-estimates from the bin duration in realtime mode (duration + 30s, min 60s), or 600s otherwise. On timeout the call returns while replay may still be running in the background.- Replay is rejected while the target sensor is streaming live data.
result = SensorControllerInstance.pauseBinReplay(sensor) # pause feeding
result = SensorControllerInstance.resumeBinReplay(sensor) # resume feeding
result = SensorControllerInstance.stopBinReplay(sensor) # abort; the blocking replayBinFile call returns
# Each returns "OK" on success or an error string otherwise.Use def parseBinToCsv(self, bin_path: str, csv_path: Optional[str] = None) -> str to convert a recorded bin file to CSV offline (parsing runs through the real pipeline; row timestamps come from the bin records):
csv_path = SensorControllerInstance.parseBinToCsv("d:/temp/test.bin")
# or with an explicit output path:
csv_path = SensorControllerInstance.parseBinToCsv("d:/temp/test.bin", "d:/temp/test.csv")
# Returns the CSV file path.CSV format — header row:
timestamp,mac,type,raw_hex,data_type,sample_rate,channel_count,lost_count,samples_info,first_sample
Row kinds in record order (config records produce no rows; a bin without a config record yields raw rows only):
rawrows — one per data record:timestamp= ISO 8601 (local time) of the bin record timestamp,mac(empty for records before the first config record),type=raw,raw_hex= raw packet bytes as hex; remaining columns empty.cmd_send/cmd_recvrows — one per command record:type=cmd_send/cmd_recv,raw_hex= command bytes as hex,data_type= decoded command name (cmd_send, e.g.NTF_DATA_START) orNAME:CODE(cmd_recv, e.g.GET_FEATURE_MAP:SUCCESS); remaining columns empty. These rows are analysis-only and are not fed into parsing.eventrows — one per BLE event record:type=event,raw_hex= event name as hex,data_type= event name (connect/disconnect/stream_start/stream_stop); remaining columns empty.parsedrows — one per parsed batch emitted by the real parsing pipeline (raw_hexempty):
| Column | Meaning |
|---|---|
timestamp |
ISO 8601 (local time) of the bin record being fed when the batch completed |
mac |
Device MAC from the config record |
type |
parsed |
data_type |
DataType enum name of the batch, e.g. NTF_EMG, NTF_EEG, NTF_ACC |
sample_rate |
Batch sample rate (Hz) |
channel_count |
Batch channel count |
lost_count |
lostPackageCount of the batch (non-zero only for new-EMG devices) |
samples_info |
Per-channel sample counts as a Python list string, e.g. [32, 32, 32] |
first_sample |
First sample of the first non-empty channel: data=<v>|raw=<raw>|imp=<impedance>|sat=<saturation>|idx=<sampleIndex>|ts=<timeStampInMs>|ch=<channelIndex>|lost=<isLost> |
setLogPath sets the SDK log directory (it must be a directory). All default file outputs live in it: the controller log, the default per-profile logs (DEBUG_LOG_PATH=True) and the default bin exports (DEBUG_BLE_DATA_PATH=True).
The controller log (sensor_controller_log_YYYYMMDD_HHMMSS.txt) holds all common logs (scan, connection management, dongle/backend, and logs of profiles whose profile log is not enabled). It is created automatically in the log directory when setDebugEnabled(True) is called, and closed on setDebugEnabled(False). Each profile log ({DeviceName}_log_YYYYMMDD_HHMMSS.txt, enabled per profile via setParam("DEBUG_LOG_PATH", ...)) contains only that profile's logs plus the bleak/bumble logs related to its current connection.
Records emitted before a log file is first created are held in a bounded memory buffer and replayed into the file on creation, so early (scan/connect) logs are not lost. The log directory and debug switch are automatically shared with the BLE subprocess, so both sides append to the same files.
SensorControllerInstance.setDebugEnabled(True)
# enable SDK debug logs; automatically creates the controller log in the log
# directory. setDebugEnabled(False) closes it.
SensorControllerInstance.setLogPath(True, "d:/temp/sdklogs")
# set the log directory (must be a directory; created if missing, rejected if
# it points to an existing file). setLogPath(False) disables file output
# (controller log, default profile logs and default bin exports).
SensorControllerInstance.setLogPath(True)
# reset to the default log directory (~/Documents/sensorsdklog)