Final project — Water Intake Tracker System

Short summary

• Simple, low-cost hydration tracking system without flow sensors or weight scales.
• Uses only motion/orientation information accompanied with some filters to estimate the volume of water per sip.
• Useful for fitness, health tracking, or simply as a learning project combining sensors, signal processing, and embedded UI.

Key features

• Detects sip events using angular velocity (gyroscope) cues.
• Estimates per-sip and cumulative volume.
• BLE for live updates or logging.
• Mobile application to record the cumulative volume and past drinking events
• Remind the user to drink water if no water intake has been made within a certain interval

Results

• 5.8% error rate compared with the real weight volume per sip

---

Details

How it works (overview)

1. Sensor & orientation

   • LSM6DSL, an IMU in the STM32 borad we use, provides gyroscope (angular velocity) data.

2. Sip detection

   • A typical sip is a forward tilt (pitch increases) followed by a return to upright.
   • The gyroscope detects angular velocity, we calculate the overall amgular velocity by the square root of y-axis and z-axis
   • Collects the forward tilt via Riemann integral of the angular velocity and check whether the tilt surpasses the threshold.
     • Mark as a sip event when the value of the current angle is larger than the threshold.
   • Filtering (low-pass) reduces noise.

3. Volume estimation

   • Estimate the drink rate based on the current angular (conducting experiemnts on various people).
     • When the angular is small, drink rate is smaller
     • When the angular is large, drink rate will be larger

4. Post-processing

   • Aggregate per-sip estimates to track total daily intake.
   • Provide real-time feedback (mobile app) via BLE.

5. UI

   • Flutter based mobile application
   • Use flutter_blue_plus as the package to connect to the hardware device via BLE
   • Display thet daily water consumption
   • Record the water consumption to the database
   • Reminds the user to hydrate after a custom period of time of no water intake

Hardware

• Microcontroller:
  • B-L475E-IOT01
• IMU:
  • LSM6DSL
• Others:
  • BLE...

Notes on sensor placement

• Mount the IMU to the side of the cup or lid in a fixed, rigid way so motion of the cup corresponds to the sensor orientation.
• Keep the IMU aligned (document axis orientation) so that the sipping event only reflects to the y-axis and the z-axis.

For more information, check out our slides

---

Improvement

Calibration & tuning

• Performing calibration and tuning for each cup, each gyroscope, or even each person will get a better estimation.

Machine Learning involvement

• Using ML-based estimation from a sequence of angular may result in a better estimation

---

Thank you for checking out ES_lab — happy building and stay hydrated!