a

Lorem ipsum dolor sit amet, consectetur adicing elit ut ullamcorper. leo, eget euismod orci. Cum sociis natoque penati bus et magnis dis.Proin gravida nibh vel velit auctor aliquet. Leo, eget euismod orci. Cum sociis natoque penati bus et magnis dis.Proin gravida nibh vel velit auctor aliquet.

  /  Project   /  Blog: What is Cognitive Workflow

Blog: What is Cognitive Workflow


Go to the profile of zia
Intelligent Workflow — A simple use case covering 5 emerging technologies (IoT, Blockchain, Machine Learning, Face Recognition, Natural Language Processing)

What is Cognitive Workflow

Cognitive Workflow is based on cognitive computing. More about cognitive computing can be found on http://bit.ly/1PBVs9C

Applying Cognitive Workflow for real world problem example

Water leaks are a common problem, if someone is available at home they may be able to shut off the valves or apply some tape around pipe and prevent leakage.

But what happens if no one is at home or it’s an REO property.

Sometimes problem may be simple enough it can be fixed by just applying tape or a patch. In other situations you may need to call up plumber for help.

For REO properties it might be lengthier process an approval might have to be taken from lending organizations or even they may have to reach out to Insurance Company. A coordination plus lot of documentation may need to be produced by maintenance companies.

Imagine property was listed and have to be taken off due to repairs. This seems to be loss of business opportunity plus loss to property and repair cost.

Not maintaining temperature greater than 55F can cause Frozen Pipe and that could potentially lead to Pipe Burst.

Let’s address this problem by applying some of the Cognitive Computing Techniques.

A water flow IoT sensor can detect pipe flow rates ranging from 0.15 to 60 liters/minute. The system can report pipe flow measurement data regularly, as well as send automatic alerts if water use is outside of an expected normal range. This allows you to identify the location of leaking pipes and prioritize repairs based on the amount of water loss that could be prevented.

Reduce costly water damage using IoT leak detectors with automated shutoff valves;

Prototype Sketch

Prototype Solution

1. IoT sensors are employed on pipes to monitor the temperature and if there is a fall in temperature beyond threshold or water leakage occurs. Sensor data is recorded and predictive analytics are executed using machine learning that determines possibility of repair required or not. Sensor data is sent using MQTT protocol.

1.1 For the sake of prototyping we can simulate IoT sensor which will generate random temperature at regular intervals using Node-Red. If temperature falls less than 32F we consider possible water leak.

2. Machine learning predictive analytics are executed on sensor generated data and will determine if repair is needed or not. If repair is needed same information is passed to block chain network.

3. Block chain network consists of 3 participants

  • Property Lender / Owner
  • Insurance Company
  • Maintenance Company

A smart contract is defined on receiving temperature data from a registered devices automatically estimated repair amount & type is calculated by Maintenance Company.

Based on repair type Insurance company will derive if repair type is covered and to what extent policy covers, estimated deductible, estimated coverage’s are derived.

All this information is sent back to Property Lender / Owner for approval.

Based on owner approval system is programmed to automatically assign servicing technician based on availability.

4. Access to Door can be granted for incoming technician using Face verification or any other security mechanism by generating PIN.

5. On opening of door by technician a welcome message would be said by BOT and problem statement will be briefed.

Technical Specifications

Hardware: Raspberry Pi, Raspberry Pi Touch Screen, Speakers, Internet connectivity is must

IoT Sensors: Water Leak Detectors, Door Lock Sensors or can be simulated (only for proof of concept)

OS: Raspberry Pi (preferred) or Windows 10 IoT or any Linux based os like Ubuntu

Software: Node-Red, MQTT

Machine Learning Tools: Big ML cloud based — no restrictions you can use Tensor Flow from Google or Sage maker Amazon or Watson from IBM.

Block chain Smart Contracts: Hyperledger Fabric, Composer — no restrictions can use any other frameworks available doesn’t matter cloud based or in house hosted) (Optional)

Face Recognition & Verification: Microsoft Azure Cognitive services Rest API

Dialog Flow: Google Assistant based on Natural Language Processing

Glimpse

Source: Artificial Intelligence on Medium

(Visited 7 times, 1 visits today)
Post a Comment

Newsletter