Making Sense of your Sensor Data

Instructors: Lara Grant and Liza Stark
Date: Wednesday, April 4, 9:30-12 pm

Soft sensors are great for custom projects, but sometimes it can be difficult to get stable readings from your sensors. This workshop will be part hands-on activity and part group discussion on how we all deal with those jumpy sensor readings. Liza and Lara will go over basics like the smoothing and calibration sketch in Arduino and bring snippets of code to share. We also want to explore how to make this weakness into a strength. Materials for sensor-making will be available. Bring your data processing techniques, your fav sensors, and your Arduino! No prior knowledge is necessary.


Purpose of the workshop:

  1. Share what you have used 
  2. Go over smoothing and calibration in Arduino
  3. Graph and compare some raw sensor data and filtered data using common filter formulas
  4. Compare and contrast how filters work on different e-textile sensors. Record and document as a reference for others.
  5. Brainstorm how jumpy readings can be advantageous

Participant goals

  • Sam – Smoothing accelerometer data; already working with a bunch of things pieced together
  • Victoria – Intro to sensors
  • Emilia – Making pressure sensors; optimal reading
  • Sophie – Reliable pressure sensor data over time and in real time; data processing for analog sensors; scaling up – working with multiple sensors
  • Taeyoon – Ways of mixing digital and analog sensors; other tricks for textile
  • Erica – See the data that you want vs how to see in a certain range

Participants shared ways of processing sensor data:

Sophie Skach – Processing data from embedded e-textile sensors in pants to identify gestures and motion.

Anastasia Pistofidou – The project TOMO uses Swept Frequency Capacitive Sensing (SFCS) a concept explored by Disney Research Labs on this paper.

Plotting sensor data

Using the Arduino Serial Plotter

Plotting in Processing and Max/MSP


Recording your sensor data

When recording sensor data while testing each filter it’s useful to record the follow:

Sensor Type | Active Material | Resting Value | Active Value | Screen Capture of Plotted Raw | Screen Capture of Plotted Filtered



Included in the Arduino IDE


Smoothing with added raw value variable to compare to average


10 sample window with 1 ms delay


50 sample window with 10 ms delay


  • Calibration – isn’t a filter, automatically sets minimums and maximums






Exponential Moving Average – Video



Low-Pass Filter 

Source for circuit

RC Low-pass filter circuit


Source for sketch


High-Pass Filter 

Source for circuit

RC High-pass filter circuit


Source for sketch


Kalman Filter library by Denys Sene

Download from GitHub and install

Kalman filter

  • Prediction step
  • Estimation step




Filters Arduino librarydownload

  • Low-pass
  • High-pass

Leave a Reply