Team:ZJU-China/Hardware/Device

"

Device

Main Device

The function of our main device is to monitor some parameters of environment, and serve as the terminal of our whole device system. It can detect the temperature, humidity, rainfall, soil moisture content, light intensity, UV intensity, TVOC and other relative parameter in real time. What’s more, there is a built-in pump inside, so our device can water the plants according to the soil moisture, or release inducers such as DAPG according to plants' status of health.

We utilized used computer case to make the shell of our device so as to make it more beautiful.

Fig.1 A photo of our device

All data will be shown in a built-in screen, and be transformed through long-distance 2.4G data transmission module.the transmission distance can be more than 2.5km, so there will be no problem if there is no network in farmland. Meanwhile, there is another built-in 2.4G module to connect with VOC device. What’s more, our device is also possessed of several extendable electric relays to connect itself to more actuators, so as to achieve more goals.

Fig.2 A photo of some components of our device

The component and cost of our device as shown in Table1.

Table1:The component and cost of our device.

ComponentCost($)
arduino mega 256010.8
arduino sensor board3.1
Temperature and humidity sensor0.9
Rain sensor0.5
UV sensor7.7
Light sensor0.9
2.4G module5.4
small 2.4G module1.8
soil moisture sensor0.9
TVOC sensor1.8
buzzer0.9
LED light0.5
water pump1.8
Relay * 31.8
LCD screen3.1
Total cost42.0

We also construct a webapp. The data we acquired can be uploaded to webapp, so that we can check the paramenters of environment and plants’ condition from mobile device in real time. You can open any browser to view the webapp without downloading any extra application.

Fig.3 A screenshot of our webapp

Slave Device

When tobacco suffers from some phytopathogens, it will release a series of Volatile Organic Compounds (VOCs), which can lead to a change of VOC composition around the plants[1][2]. We utilized 10 highly sensitive Complementary Metal Oxide Semiconductor(CMOS) gas detectors to capture this change, using the machine learning method to analyze the obtained data, and establish the training set in advance. In this way, we can distinguish the health state of tobaccos according to the air components data.

The sensitive material of our gas detectors is some highly active metal oxide semiconductors, such as SnO2. Under the working conditions, when the sensor encounters a reducing gas, the oxygen anion on its surface decreases due to the redox reaction with the reducing gas, resulting in a decrease in the resistance of the sensor. The resistance of the sensor is related to the concentration of the gas, and each sensor has a special sensitivity to a certain type or a class of gas. It is important to note that e-nose can't sense a specific gas component like mass spectrometers did, instead, it catches the overall characteristics of VOCs as a "fingerprint".

Highly sensitive CMOS gas detectors we used are illustrated in Table.2.

NumberSensor TypePerformance characteristics
aIST-8000Highly sensitive to all types of VOC
bTGS2600Sensitive to cigarette smoke and cooking odors
cTGS2610Sensitive to alkanes such as liquid gas propane butane, low sensitivity to alcohol
dTGS2603Sensitive to ammonia and sulfide gas
eMS1100Highly Sensitive to aldehydes, toluene and organic solvents
fTGS2611Sensitive to methane
gTGS2602Highly sensitive to all types of VOC
hMQ-7Sensitive to carbon monoxide and other gases
iMQ-135Sensitive to ammonia,sulfide and benzene vapor, or harmful smokes
jTGS822Sensitive to alcohol and organic solvents
k(substitute b later)iAQ-coreExtremely high sensitivity to all types of VOC and can output the equivalent concentration directly

Table.2 Highly sensitive CMOS gas detectors we used(All of these detectors have long-term stability)

Fig.4 A photo of our VOC device

We also constructed the relevant gas path. We utilized the three-way valve and SCM(Single Chip Microcomputer) control system to achieve automatic and standardized measurement, so that the data between the groups can be compared. Measured data can be directly transmitted to the computer through a small 2.4G data transmission module, or be sent to the main device to achieve ultra-long distance transmission. When the slave device is around the main device, the main device will automatically recognize the presence of it and begin data transmission,at the same time a green LED will be on to show the connection status. The measurement diagram and flow chart are as Fig.5 shows.

Fig.5 Process

We built median filter algorithm in arduino SCM to remove some outliers. Then, we preprocessed the raw data after getting them: First, we identified and removed the base line value; Then, we read the response value on 1min, 2 min, the integral average value and maximum value of the response curve. Then, we used the machine learning method to analyze the obtained data. Click here to see the details of our modeling process.

We achieved more than 85% accuracy rate when sensing the tobaccos’ health condition. Moreover, based on the result of the modeling, four CMOS sensors were enough to make a judgement for tobaccos’ health condition, which means that we can further reduce the cost of our device.

We have designed a visual interface, on which you can choose different work mode of our device:

Fig.6 Photos of our device's interface

What's more, we have constructed a webapp. You can see the current data and the line chart of some recent data on it. We plan to add more functions to it in future, for example, we can get to know whether our plants are in health condition directly

Fig.7 Some screenshots of our webapp



How to use our device? Let's watch the following video!

But the sensitivity of our device is still limited when facing real world conditions. It is hard to fit different conditions in different field. Therefore, we come up with a plan to solve this problem. Click here to see our improvements.

Reference

[1]Aksenov A A, Novillo A V G, Sankaran S, et al. Volatile Organic Compounds (VOCs) for Noninvasive Plant Diagnostics[J]. Acs Symposium, 2013, 1141:73–95.

[2]Dicke M, Loon J J A V, Soler R. Chemical complexity of volatiles from plants induced by multiple attack[J]. Nature Chemical Biology, 2009, 5(5):317-324.