Greenhouse status monitor

Nowadays the system development is oriented to services and platforms like SmartEdge Agile and Brainium simplify the process

Inspiration

Some regions of Mexico work with greenhouses. Many times the users need data in order to make decisions about irrigation or only monitor the conditions in the greenhouse to work securely. Acquiring data we can understand what is happening in the greenhouse and guarantee the health of the workers as well as harvest. There are solutions only for monitor some parameter about external greenhouse weather but not for the interior. With the acquired data the users can estimate if the conditions in the greenhouse are secure for the human work at specific hour of the day. Sometimes the high temperature and humidity can affect the health of the workers and the plants development. The data catched by the device will send to a monitor terminal as a PC, because some users only know the basic technology. According the data, the leader can say if the condition into the greenhouses is safe.

The Setup

In order to test the platform there are three principal components: A gateway, the network infrastructure provide by a Lenovo Stack and the SmartEdge

Building

In the first step, the connection and test of the device is needed. To pair the device, one need a smartphone with the brainium app installed, that allows to the smartphone acts as gateway and, by brainium platform, add our device and configure it. In addition, we should create a project to use the device and start an infrastructure test.

In the green house we need consider an inportant factor that the SmartEdge shares with other wireless devices, the installation should be far of a metal structure like pole, avoid this could afect the connection to gateway.

As the title says, the project's approach is for monitor the greenhouse weather parameters. So, We need add those parameters that are important to determine if the inner ambient is secure to work by an external auditor.

When we add a tracking widget, the show data format and the device rate could be configured according our needs of visualization. In addition, alarms can be configured to alert out of parameters, an example added is the infrared lightness to monitor the sunlight impact in the green house. Other parameters added were visible lightness, humidity, temperature, and battery state, consecuently, our data show layout is the next.

in the left, the user can see the alerts of out parameters, in the right we have other parameters that allow observe the behaviour of the greenhouse and determine if the parameters are safe to work. The auditor can take decisions about if the workers continue their activities or suspend by security.

An interesting feature of SmartEdge is the AI edge where can generate models according to the data. An example is motion detection, here we can sample gestures to train a model an alert about an event as fall of the device of men requesting help in the greenhouse. The platform allow to sample, train, and implement the model. Other important feature is that the model is described by its training features, accuracy, f-measure and so on.

Final remarks

Because the SmartEdge reduce the sensing and AI infracstructure, we have an excellent device for AI in the edge.

It is a powerfull platform but I faced with some inconveniences like baterry charge use time, I do not know if the device cosumes a lot of power by the first use, the battery does not charge correctly or firmware bugs.

We should know the AI approach to use the sampler and trainer for model generation in order to reduce the work time needed to implement predective maintenance or motion recognition.

Results

According to data collected and workers behaviour, we got some rules to handle greenhouse human resources. The better schedule for work is from the first morning hour, like 5:00 or 6:00 a.m., to the 11:00 a.m., after this time the people is considered over exposed and it has sunstroke, avoids liquids and foods during the work, and a slept behaiviour because the high temperatures into the greenhouse. The light sensor achieves a

In adition, the crops need an optimal temperature, the device acquire the temperature data and an alarm is set to manual control of the ventilation, this according to the infrastructure available in the greenhouse. This action helps to control the humidity level too in the greenhouse but we need know the environmental conditions to get a balance.

The proximity sensor allows sense the presence of someone in the greenhouse but is not efficient at all because the people's route changes according to the activities like watering, grass removal and so on.