In agribusiness, biosecurity is crucial, since diseases that may not be dangerous for humans are fatal for pigs.
In order to prevent unwanted diseases from entering the herd, all of the farm workers must comply with the biosafety policy that includes taking a shower. Unfortunately, not all people take biosecurity measures seriously and avoid a shower room.
Our client - an agricultural company that specializes in growing pigs and cattle and selling self-produced food products - wished to design a solution that will monitor the compliance of biosecurity measures.
As the client said:
Our engineering team observed the farm biosafety area, which includes two changing rooms and showers.
To fulfill customer expectations, our experts created an indoor location tracking system that shows in which of the rooms the employee is located and how much time they spend there. Knowing how much time they spend and how intensive are their movements in a shower room, we may monitor the compliance of biosecurity measures. Also, the proposed solution would remove any human factors from the hygiene control process.
The solution was achieved with the help of a beacon device, smartwatch, and remote server for computation.
We had two options:
- Beacons on walls
- Beacons on employee (this option is more expensive regarding setup, and maintenance; and worse regarding area scaling)
The challenges we faced during all the development stages made us go through water and fire. But it was definitely worth the effort.
Continue reading to know what may stumble you on your way of developing a hygiene compliance monitoring system and how to overcome these difficulties.
One of the cons of the smartwatches is that they do not run for long, therefore, we can’t provide the instant transmission of data. Initially, we tried to divide the transmission of data into small portions. But being connected to the WiFi the smartwatch battery went dead really fast.
Our next idea was to transmit data to the server only when smartwatches are charging. And this solution turned out to be successful. With this solution, smartwatch battery life lasts for 8 hours which is also the duration of work shift.
Imagine taking a shower. It’s rather an active process, right? You may be turning around, moving your hands in different directions. And that’s fine unless you are not wearing a smartwatch that is receiving the beacon signal, because these moves cause beacon signal loss. What is more, the smartwatch may catch signals of other beacons.
These actions got reflected on data. Analyzing graphics, we noticed that there occur discontinuous jumps that distort the accuracy of data. To solve this issue, we used specific algorithm methods and data filtering. Data filtering discarded unnecessary jumps and left only the data that was needed to calculate the duration of stay in a room.
Next challenge occurred during the replication of a demo on the farm. It turned out that the walls are too thin for the beacon signals. Since beacon signals break through thin walls, the RSSI values of the 3 rooms differ only on a few signal units.
We tried to solve this solution with such options as paint with silver ions, carbon fabric, and steel sheets. After several tests, we obtained the best result from a steel sheet. Why?
Because, the steel sheet acts as a screen to amplify the signal of the beacon and does not allow the signal to break through the wall on which the beacon is placed, therefore suppressing the signal for other beacons.
As you may see on the photo, standing near the first beacon, the smartwatch received the strongest signal from the first beacon respectively. Signal received from the second beacon is a way lower, and from the third is totally absent.
The correct location of the beacon on the steel sheet also matters, as it was found that the water was disrupting the signal of beacons.
To determine the needed location of a beacon, we used the formula λ/4, where λ is the wavelength. In our case with 2.4 GHz, the wavelength is 12.5 cm.
L = 12.5/4 = 3.1
In the middle of the sheet, we placed a 3.1 cm thick gasket and attached a beacon. This technology has improved the signal of the necessary beacon and reduced the beacon signals from other rooms by approximately 15-25 units.
But challenges with the walls did not end here.
Development Challenge #5: Smartwatch RF signals interfere with one another outside the changing room
Walls outside the biosafety area are as thin as the walls inside the area. That means that people that walked in a hall with their smartwatches were receiving the beacon signal.
We instantly rejected the option with steel sheets, since an area we needed to cover was over 50 square meters - it would be a time-consuming procedure that limits access to these rooms for a long period of time. Therefore, we decided to dig deeper.
We found a German carbon paint HF65 which has the properties of reflecting the radio signal.
After several tests conducted in the office, like creating a box that represents a model of a room. In the tested version, drywall and one coat of paint were used.
In real conditions, the existing paint will be applied in two layers to ensure an even better result, ensuring the complete absence of signal.
Human factors have also played a role. The process of debugging was prolonged since the farm workers needed to get familiar with the devices. There were also some special conditions, due to the human nature of forgetting things.
To prevent smartwatches from being uncharged and left somewhere else, a custom smartwatch charging station was designed. It also serves as a point of smartwatch storage.
The process of charging takes place with the use of a magnetic end of the cable that snaps into place to the charging port when it’s properly connected to the smart watch.
Once we finished maintaining the hardware, our team proceeded to the data science issues. The Lemberg team of data scientists had to design scenarios of data collection during the actual showering procedure. Gathering of test data was conducted through various ways. A three-people focus group gathered data at home, some of the employees simulated showering at the office, as well as we continuously gathered showering data at the farm.
All these data sets were labeled correctly and ready for further modeling by Lemberg data scientists. The resulting motion analysis model was reflected on a dashboard - a platform for at-a-glance viewing of showering analytics.
Building a comprehensive hygiene compliance monitoring system would be impossible without a thorough analysis of collected data.
Analyzing the data, we understood that all people have their own routine of taking a shower. Understanding the patterns of human behaviour is the next challenge our team is working on.
Knowing those patterns, we would be able to estimate how properly a person is taking a shower and, consequently, how strictly the biosecurity measures are met.
Right now, we are trying to write down different scenarios for different people in order to see how this data would be reflected on the server. But are still in progress with this solution.
Developing is challenging. You may read a lot about embedded software development services, but the experience you gain from one and a half years of developing your own one is worth its weight in gold. Most of the challenges occur during the very process of development or after product release.
Only putting your shoulder to the wheel, you may notice things that don't fit, employees that don’t use your device the way you expect it, different signals range, and collected data that doesn’t fit your assumptions. 1.5 years give you a clear understanding of certain hardware setup pros and cons; and how device specifications (like 10 hours battery life) change after 1 month of usage, and what you may expect in 12 months.
Here at Lemberg, we know a lot of the nuances of the development of data science and IoT solutions. Do not hesitate to discuss the prospect of our partnership with Slavic Voitovych. He is always ready to have a profound talk with you.
© Lemberg Solutions, 2020