The document discusses SensorTile.box, a development platform from STMicroelectronics for exploring sensor applications. It provides an overview of SensorTile.box and its sensors, as well as use cases and the development ecosystem. The platform allows progression from entry-level use to expert and professional modes with more customization options. Vibration monitoring is presented as an example application, showing how it can be implemented and optimized at different levels.
6. Play with Sensors:
Log data and
explore new sensor
functionalities
Proof of Concept:
Test and validate
your ideas by
customizing the App
Rapid Prototyping:
Customize, write
code and make it
yours
Get Familiar with Sensors and Speed Up your Development
11. Inside the SensorTile.box
Sensing, Processing and Connectiviy
11
Low-power 6-axis IMU, with
Machine Learning Core
LSM6DSOX
High-performance and low
power accelerometers
LIS3DHH & LIS2DW12
Magnetometer
LIS2MDL
Altimeter / Pressure sensor
LPS22HH
Accurate temperature sensor
STTS751
Humidity sensor
HTS221
Analog wide-band microphone
MP23ABS1
Motion Sensors Environmental Sensors Processing
STM32L4 low-power MCU
STM32L4R9ZIJ6
Connectivity
Bluetooth Low Energy Module
SPBTLE-1S
Power management
Battery charger
STBC02
12. Entry Working Mode
No Need to Program Anything
12
Entry
Mode
Human Activity
Recognition
Pedometer
Sensor Fusion
Quaternion
Compass And
Level
Vibration
Monitor
Data Recorder
AI based
In-Vehicle
Baby Alarm
Baby Crying
Detector
Barometer
13. Expert Working Mode
Expert
Mode
• Define additional applications, like in Lego bricks
• Build his own application, without programming
6
14. Pro Working Mode 14
Pro
Mode
Full compatibility and support of STM32 ODE
Professional Level of Development
FP-SNS-STBOX1 AlgoBuilder
Unicleo GUI
16. Vibration Analysis – SensorTile.box 16Machinecondition
Time
Conditions
start to
change
Vibration
Noise
Heat
Power
Smoke
Failure
minutes
days
months
weeks
From Entry to Pro Mode Experiences
17. Entry Mode: Vibration monitoring
Vibration monitoring: FFT transform of data from wideband accelerometer
Enables predictive maintenance
• 1st step is training
• Record typical/normal data on SD card
• 2nd step is compare
• Compare and turn-on LED if atypical/abnormal (this can be the early warning)
17
18. Vibration monitoring - Training 18
1. Press the button
Create apps
2. Vibration Monitor
(Training)
3. Touch
to upload
4. Touch
to upload
19. Vibration monitoring - Training 19
1. Press the button
Connect to a device
2. Select your
SensorTile.Box
3. Touch
Start Logging
4. Touch
Stop Logging
Setting baseline vibration
20. Vibration monitoring - Training 20
1. Connect via USB, select USB drive
and copy gold_vibration_pattern.csv
2. Get CSV data, open with
Excel or any text editor
FFT LOG ON SD CARD
21. Vibration monitoring - Compare 21
1. Press the button
Create apps
2. Vibration Monitor
(Compare)
3. Touch
to upload
4. Touch
to upload
22. Vibration monitoring - Compare 22
1. Confirm it is
Ok
2. Wait for the upload
to be completed
3. On confirmation
touch Ok
4. Go back
main screen
23. Vibration monitoring - Compare 23
1. Press the button
Connect to a device
2. Select your
SensorTile.Box
3. Typical/normal
LED is OFF
4. Atypical/abnormal
LED is ON
Vibration
OK
Vibration Outside
Baseline
24. Vibration monitoring - Compare 24
1. Typical/normal
LED is OFF
2. Atypical/abnormal
LED is ON
3. Start and stop logging,
get CSV data
25. Entry Mode → Expert Mode 25
Creating a customized Vibration Monitoring Application
Key Customization Features in Expert Mode:
• Select desired input → Wide bandwidth accelerometer (vibrometer)
• Select functions → FFT transform (256 points)
• Select outputs → Stream to Bluetooth
• Save customized application
26. Vibration Monitoring App 1/4 26
1. Press the button
Create apps
2. Scroll down, touch
Expert view
3. Touch
New app
Run the ST BLE
Sensor app.
4. Touch to select
an input
27. Vibration Monitoring App 2/4 27
1. Select Vibrometer,
then Set Input
2. Touch the gear
to change settings
3. High perf, 50Hz,
HPF, Save Config
4. Touch to select
a function
28. Vibration Monitoring App 3/4 28
1. Select
FFT
2. Touch the gear
to change setting
3. Select FFT length,
then Save Config
4. Touch to select
output method
29. Vibration Monitoring App 4/4 29
1. Stream to BLE,
then Continue
2. Touch
Save App
3. Set name,
then Finish
4. The new app
is READY!
Vibration Monitor
Vibration Monitor
30. Vibration Monitoring App - Evaluation 1/3 30
1. Touch
to upload
2. Touch
Ok
3. Wait for the upload
to be completed
4. Touch
Ok
31. Vibration Monitoring App - Evaluation 2/3 31
1. Touch to go back
to main screen
2. Touch
Connect to a device
3. Select your
device
4. Vibration
trial
32. Vibration Monitoring App - Evaluation 3/3 32
1. Low frequency
vibration
2. Higher frequency
vibration
33. Expert Mode → Professional Mode 33
Full compatibility and support of STM32 ODE
Key Features in Professional Mode:
• STM32 Programming
• Dedicated STM32 Cube Function Pack – FP-SNS-STBOX1V1
• AlgoBuilder
• UnicleoGUI
34. AlgoBuilder 34
An application for the graphical design and testing of algorithms
Existing algorithms
User-defined data processing
blocks
Additional functionalities
GUI quickly elaborates proto applications
for
MEMS sensors
and
STM32 microcontrollers
AlgoBuilder
eases the process of
implementing
proof of concept
without writing the code
36. Unicleo GUI
• Displays data from connected sensors (time
plot, scatter plot, 3D plot)
• Saves data to tab separated (TSV)
or comma separated (CSV) files
• Program Nucleo with selected
MotionXX middleware library
• Reads from and writes to registers
of sensors on X-Nucleo boards
• Connects over USB or BLE
• Supports many platforms
• Nucleo-L476/F401/L152/L073 with
X-Nucleo-IKS01A2/A3 or IKS02A1
• BlueNRG-Tile
• SensorTile.box
36
38. AlgoBuilder – Vibration Monitor 38
3. Click OK
2. Select Target NUCLEO-L476RG + X-NUCLEO-IKS01A3
1. Select the desired compiler, in this example: STM32CubeIDE
39. AlgoBuilder – Vibration Monitor 39
1. Place [Sensor Hub] to workspace
2. Change Accelerometer Full Scale to 4g
3. Output Data Rate ODR 100Hz
40. AlgoBuilder – Vibration Monitor 40
1. Add Acceleration [g] function block
2. Connect the blocks
41. AlgoBuilder – Vibration Monitor 41
1. Add Graph function block
3. Connect the blocks
2. Set properties, change default values:
• Number of Curves:3
• Graph Name: Acceleration
• Waveform1: X axis
• Waveform2: Y axis
• Waveform3: Z axis
42. AlgoBuilder – Vibration Monitor 42
1. Add Demux Float function block from the Other library
2. Set properties 3 outputs
3. Connect
43. AlgoBuilder – Vibration Monitor 43
1. Add 3x FFT function block from the FFT Library
2. Set properties 256 data length for the 3 blocks
3. Connect the blocks
44. AlgoBuilder – Vibration Monitor 44
1. Add 3x FFT Plot function block from the Display Library
2. Add to Plot Name the axes FFT plot X, Y and Z
for each block
3. Set properties 256 data length & Full Scale 1 for all blocks
4. Connect the blocks
45. AlgoBuilder – Vibration Monitor 45
1. Add 3x FFT PEAK function block from the FFT Library
2. Set properties 256 data length for all the blocks
3. Connect the blocks
46. AlgoBuilder – Vibration Monitor 46
1. Add MUX float function block from the Other Library
2. Set properties 6 inputs
3. Connect the blocks
47. AlgoBuilder – Vibration Monitor 47
1. Add Value function block from the Display library
2. Set properties 6 values and the labels
3. Connect the blocks
54. FP-SNS-STBOX1
• BLEMLC Machine Learning Core demo, config by Unico GUI
54
Simple application example using the MLC (Machine Learning
Core) embedded in LSM6DSOX, which also embeds the FSM
(Finite State Machine).
The Unico GUI is the tool to generate the code to configure the
MLC in the sensor. A third party tool is used to design the decision
trees. – see BLEMLC details
55. LSM6DSOX Smart Sensor 55
Sensing &
pre-
processing
High-level
processing
Sensing
Full activity
monitoring
processing
Power
optimization
at system
level
Finite State Machine Machine Learning Core
NEXTRESET
COMMAND
PARAMETERS
CONT
node
nodenode
node
Start node
node
The first Smart Sensor embedding
FSM and MLC allows sensors to process data with reduced help of a host Microcontroller
Start
NEXTRESET
NEXTRESET
56. BLEMLC: Application Example 56
Using the
ST BLE
Sensor
app.
1. Select
Activity Recognition
2. Try different
Activities
3. Display activity
in real-time
57. Design Flow
Tutorials
57
Step by step guide on YouTube,
STMicroelectronics channel
• LSM6DSOX
• Introduction to design flow
• Data collection
• Labeling and features
• Decision Tree generation
• Register configuration
58. More MLC examples from ST 58
https://github.com/STMicroelectronics/STMems_Machine_Learning_Core
59. FP-AI-SENSING1 59
• The package enables advanced
applications such as human activity
recognition or audio scene classification, on
the basis of outputs generated by neural
networks (NN)
• The NN are implemented by a multi-
network library supporting both floating and
fixed point arithmetics, generated by the X-
CUBE-AI extension for STM32CubeMX
tool.
STM32Cube function pack: IoT node with artificial intelligence (AI) application
60. FP-ATR-BLE1 60Coming Soon
• The FP-ATR-BLE1 is STM32Cube Function
Pack for low power asset tracking using
BLE connectivity for the SensorTile.box
Key features
• Complete sample application on how to
create one asset tracking application
controlled by BLE connectivity
• uSD DataLogging
• Wake-up, Tilt and Orientation detection
by the on-board accelerometer
Pro Mode: Asset Tracking Function Pack
62. Entry Mode:
Log data and
explore new sensor
functionalities
Expert Mode:
Test and validate
your ideas by
customizing the App
Professional Mode:
Customize, write
code and make it
yours
SensorTile.box Hardware
64. ST Community – Mass Market Support
For more information on sensors: www.st.com/sensors
Information on longevity: 10yr Longevity Program
Android / Linux / Open Drivers: Drivers for MEMS
Mass Market Support: Online Support
64
• YouTube Videos:
65. Takeaways
• Internet of Things offers many opportunities for ST
from Sensor to Cloud
• ST has all the building blocks for the IoT
• Our Sensors portfolio brings high performance devices from
consumer to industrial applications
• SensorTile.box jumpstarts your understanding, evaluation and
development of sensors targeting IoT and wearable
applications
• We keep the commitment on MEMS innovation
• Our constantly expanding development ecosystem
makes design fast
65