IoT-Based Fall Detection Prototype Using MPU6050 and Threshold-Based Motion Analysis

Abstract

Falls can cause serious injury when assistance is delayed. This paper presents a hardware demonstration prototype using an ESP32 DevKit V1 and MPU6050 motion sensor to identify motion patterns associated with a possible fall. The implemented algorithm applies sequential thresholds to resultant acceleration and gyroscope angular-velocity magnitude; no trained machine-learning model is claimed. Hall-effect and capacitive inputs gate monitoring. When the programmed conditions are satisfied, the controller updates an LCD, activates a buzzer, checks for a recent valid GPS location, and attempts SMS transmission through an A7670C cellular modem. Functional development tests observed ESP32 serial output, Wi-Fi connection, GPS data, fall-sequence status, SMS submission, and an SMS-received indication. Detection accuracy, false-alarm rate, and delivery timing have not been quantified through repeated controlled trials.

Introduction

The paper presents an IoT-Based Fall Detection and Alert System designed to identify possible falls and send alerts to caregivers or emergency contacts. The system is intended to improve the safety of older adults and people who may be unable to ask for help after falling.

It uses an ESP32 microcontroller and an MPU6050 motion sensor to monitor acceleration and angular velocity. When a programmed motion pattern indicates a possible fall, the system activates a buzzer, displays the status on an LCD, obtains the GPS location if available and attempts to send an SMS alert through a cellular modem.

1. Related Work

Existing fall-detection systems use different technologies, including:

• Accelerometers to measure body movement.

• Gyroscopes to measure rotational movement.

• Computer vision to identify falling activities.

• Threshold-based algorithms to detect sudden movements.

• Machine-learning techniques to classify normal and abnormal activities.

Combining accelerometer and gyroscope readings provides better information about movement. However, sudden activities such as jumping, sitting quickly or handling the device can sometimes be mistaken for falls.

2. Proposed System

The proposed system is a hardware demonstration prototype that uses sensors and communication modules to detect possible falls and generate alerts.

Its working process is:

1. The MPU6050 continuously measures acceleration and angular velocity.

2. The ESP32 collects and calibrates the sensor readings.

3. A threshold-based algorithm analyzes the movement patterns.

4. Hall-effect and capacitive sensors check the required monitoring conditions.

5. When the programmed fall-detection sequence is satisfied, a fall alert is triggered.

6. The buzzer produces a local alarm.

7. The GPS module provides the current location if a valid GPS signal is available.

8. The GSM module attempts to send an SMS alert to the registered contact.

3. Hardware Components

4. Threshold-Based Fall Detection Algorithm

The system uses a sequential threshold-based algorithm to identify possible falls.

The MPU6050 provides three-axis acceleration (X, Y and Z) and three-axis gyroscope measurements.

The raw accelerometer readings are converted into approximate acceleration values using calibration offsets.

The main steps include:

• Collecting raw sensor readings.

• Removing sensor bias using calibration values.

• Calculating acceleration and angular velocity.

• Comparing the measurements against predefined thresholds.

• Checking whether the required motion sequence is satisfied.

• Triggering an alert when the programmed conditions are met.

The calibration offsets depend on the individual sensor, its orientation and operating conditions. Therefore, they must be verified on the actual hardware.

5. Monitoring-Gate Mechanism

The Hall-effect and capacitive sensors act as monitoring gates.

• The Hall-effect sensor checks the configured belt-state condition.

• The capacitive sensor checks the contact-related condition.

• If either required condition is absent, the system resets the partial fall-detection sequence.

• If both conditions are satisfied, fall detection continues.

Conclusion

This paper describes an IoT-based fall-detection demonstration prototype using an ESP32 DevKit V1, MPU6050, Hall-effect and capacitive inputs, LCD, buzzer, GPS receiver, and A7670C modem. The implemented logic uses sequential acceleration and angular-velocity thresholds rather than a trained machine-learning model. The reported development tests observed controller/serial output, GPS data and an outdoor location fix, trigger/fall status, SMS submission, and an SMS-received indication. These results demonstrate operation of selected system functions under the recorded conditions, but do not establish detection accuracy or communication reliability. Future work includes repeating the calibration procedure for each sensor, confirming full-scale register settings, tuning thresholds with labeled trials, adding an explicit final-posture check, measuring false alarms and missed falls, and quantifying GPS and SMS timing over repeated tests. A machine-learning method may be investigated later only after collecting a representative dataset and independently evaluating a trained model. Any future paper should report the actual test protocol and measured results, not assumed performance values.

References

[1] M. Mubashir, L. Shao, and L. Seed, “A survey on fall detection: Principles and approaches,” Neurocomputing, vol. 100, pp. 144–152, 2013. [2] A. Igual, C. Medrano, and I. Plaza, “Challenges, issues and trends in fall detection systems,” BioMedical Engineering OnLine, vol. 12, 2013. [3] A. K. Bourke, J. V. O’Brien, and G. M. Lyons, “Evaluation of a threshold-based tri-axial accelerometer fall detection algorithm,” Gait & Posture, vol. 26, no. 2, pp. 194–199, 2007. [4] TDK InvenSense, “MPU-6000 and MPU-6050 Product Specification,” product documentation. [5] Espressif Systems, “ESP32 Series Datasheet,” product documentation.

Copyright

Copyright © 2026 Dr. N. C. Patil , Manoj H, Namratha K P, Harish D, Khushboo D. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.