Y’all Ready for This??
So here is what Sentinel is about :
So basically as you know we have drainage pipes which with time develop blockages due to various reasons. These blockages right now have NO way of being detected until they cause waterlogging and cause infrastructural damage and are costly to repair. But these blockages are actually NOT UNDETECTABLE.
So here is the thing -> when the blockages start forming, the water level in the drains acts abnormally. These abnormalities can be detected by using an Anomaly Detection Model .So my entire project is based on this idea.
Here is a 3D animation I spent way too much time on.
Comments 6
How exactly does it work? what sensors does it use? How did you write the code for water anomaly, did you have data to back the code?
Great Video but I think you need to recharge the device to often instead of measuring through RADAR i would use the conductive property of water to make the rising water as a switch so the whole circuit is only powered when the water had been rising.
@weebintent So Basically Its Supposed to Be Installed in Drainage Pipes Above the Water. There Will be Nodes of these Devices Forming a Network Chain. So Basically Heres A Rough Sketch -> The First Node in the Chain Wakes up from Deep Sleep (THe State in which Almost No Battery Used) Reads the Water Level Using a water level Radar Sensor, This is then Transmitted via the Lora To the next Node in the Chain. The Next Node Wakes , Listens to the previous Node , Receives the Data, Appends its own reading and Passes it to the Next Node.
The Entire Chain is Managed by Unique Node IDs and also Since the Nodes can be Damaged Causing the Entire Network to Fail, Theres also a Backup, like If a Node is damaged then the Previous Node’s Data Will be Received by the Next Node Skipping the Broken Node Entirely.
Also Regarding the Actual AI Model, Since its Not Quite Possible to Get Actual readings Right Now, I plan to generate Data using a Realistic Env with Water Usage Patterns like High Water Usage in Morning and High water level during Rain time etc.
Then I’ll Use a LSTM Anomaly Detection Model and Train it using One-Class AI training Strategy So that the Model Can be Trained Without Needing An Equal amount of Blocked Water Level Data.
@MrKnowledge Yes Regarding that -> So Basically Each node uses Low Power Modules and Also I Am planning to Integrate DeepSleep Into the Processor and It wakes up only at a Specified time Range. Also I cannot Use your Idea Because Basically the Blockages Take Time to Form and Need Continous Monitoring to Detect those Slight Variations and Also Introducing a Component Which is In Contact with the Water Increases Chances Of Failure.
Also Right Now Im Focusing More on the Actual Model Part And am not Entirely Decided on the Hardware Config etc.
I Am open For any Suggestions! Thanks for Showing your Support!!!
Wow this is so cool. I actually did think of doing something like that since my area did flood once and all the drains are clogged and my neighbour’s houses were all wet, but I end up not doing cuz it sounds too hard. :D
WOW
Sign in to join the conversation.