WaterX – An Ingenious Water Leakage Detection Solution
Overview
Water leaks, whether small drips or major floods, can cause significant damage to properties and incur hefty repair costs. Now, Imagine a world where you never have to worry about hidden water leaks in your toilet tanks that cause damage to your property as well. WaterX is here to make that world a reality.
An exceptional creation of RevolveAI, WaterX is an ML-powered water leakage detection system designed
as a compact IoT device to tackle the water leakage problem, particularly in toilet tanks.
Timeline
8-9 months
Resources
4-5 persons
How Did Our Journey Toward WaterX Begin?
Our journey with WaterX began when a client approached us with a simple yet crucial problem: “How to detect and prevent water leaks in toilet tanks quickly? Also, when there are hundreds of washrooms.”
The idea was to make a system as a device that detects water leaks and informs the user instantly. The major modules were:
Small device that works 24/7
Generate alerts to notify users about leakage
The solution mentions the leakage level as well
The Real Challenge
Crafting an ML-powered solution for this problem was no small feat. However, our adept team persevered to create a solution that perfectly met the client’s needs.
The biggest challenge was training a model for devices with very small memory. In addition, the device’s battery only has one to two years of life. So, keeping the whole system working 24/7 with such a battery life span was quite challenging.
After detailed R&D sessions, RevolveAI harnessed the breakthrough Artificial Intelligence and Machine learning models to create a compact, battery-powered device that floats at the top of the water tank and detects its leakages.
The Process
The journey begins with collecting real-time data using microchips of accelerometers and gyroscopes. We then create pre-feature vectors, which are filtered to remove anomalies, resulting in clean feature vectors.
After applying a low-pass filter to eliminate noise, we use normalization techniques for feature engineering. Finally, we feed the feature vector into a Convolutional Neural Network(CNN) model for the ultimate prediction. We use False Alarm frequency to gauge the model’s efficiency.
Device Overview
The tiny IoT device with an ESP32 board provides strong WiFi and Bluetooth functionality, while there is an IMU sensor that tracks the water movements in the tank. Also, a device coded with Arduino enables the device to work like a pro.
This device uses a 1000mA battery. In addition, it has a sleeping mechanism that prolongs its lifespan to a year to two years. Moreover, the device leverages the CNN detection model as its machine-learning component.
How WaterX Works?
WaterX’s smart approach lets you jump into action fast, preventing costly damage and inconveniences. Here’s how it works:
How WaterX Works?
WaterX’s smart approach lets you jump into action fast, preventing costly damage and inconveniences. Here’s how it works:
The ML Pipeline
The ML Pipeline
Leakage Detection at Its Finest
With WaterX telling you the level of water leakage, you have an AI-powered guard that protects your property against the potentially disastrous consequences of water leaks.
Stumbling blocks we had
During the development of this subtle solution, we had a few roadblocks:
The Final Product
Whether it’s a minor drip or a major gush, WaterX is your vigilant partner, helping you protect your property and, ultimately, your peace of mind.
It’s a tailor-made, AI-powered solution specially designed for hotels in the USA, where a multitude of rooms makes manual leak detection impractical. While it’s not meant for domestic use, it’s the perfect guardian for massive setups.
Tools & Technologies
Our waterX system utilizes a robust technology stack with C++ and Python for efficiency, Deep Learning (CNN) and Pointnet for precision, and Tensorflow for seamless integration, ensuring reliable and timely toilet tank leak detection.
TensorFlow
ESP 32
Arduino
Embedded C
