Artificial Intelligence: Improving the Effectiveness and Efficiency of Decision Making in Agriculture
Introduction
The world has been granted the discovery and application of a technology that has drastically altered the plethora of business processes harmoniously working together to achieve a desired outcome. Artificial intelligence (AI) refers to the science and engineering of constructing intelligent machines, especially intelligent computer programs (McCarthy, n.d.); the machine consists of hardware and one or more algorithms that communicate and collectively effectuate the ability to learn or understand new and trying situations. Numerous applications involving AI have been implemented in disparate industries with work performed by a host of information technology professionals to advance the technology and the profitable use cases thereof.
The subsequent portions of this report address the methods by which AI can improve the effectiveness and efficiency of business decision-making in the agriculture industry. The subject will be analyzed with its importance justified. Thereafter evidence for the solution to the question being addressed is presented, followed by an analysis of three alternative implementations of AI to improve the effectiveness and efficiency of business decision-making in agriculture. The report is concluded with a choice and explanation of the optimal solution deduced from the content presented in the previous sections.
Problem
The proponents of AI and stakeholders involved therewith advertise the technology as proffering value to businesses in a variety of industries. The value may consist of fostered profitability and augmented efficiency in business processes. Brian Goehring and Anthony Marshall of IBM reported more than 85% of advanced adopters of AI are reducing operating costs; advanced adopters of AI attribute 10–12% points of revenue gains (or erosion offset) to AI (2020). Per the definition of artificial intelligence disclosed in the introduction, AI on its own is incapable of rendering increased profitability, efficiency, and effectiveness for an enterprise, but groups and individuals are involved in the development, implementation, and management of AI to reap the benefits.
Agriculture consists of business processes that generate data that can be used to improve the effectiveness and efficiency of decision-making. There are factors beyond the scope of agricultural business processes that generate data relevant to the decision-making process a person in the industry performs. Acquiring and judiciously implementing the external data with the data generated by business processes can provide numerous advantages. Due to the natural limitations of man and the lack of time, much of the benefits that are available for business decision-making that result from the collection, storing, processing, and presentation of data are lost or not utilized as efficacious as an AI system working with a human would use it to optimize decision-making. There are tasks relevant to agriculture that may be automated; the lack of such AI implementations adversely affects operational effectiveness and efficiency. Such circumstances if not improved may abate the yields achieved by farmers, thus adversely affecting the business. The financial and operational detriment that people in the agriculture industry experience due to the lack of AI implementation may be solved given the current state of technology.
How can AI improve the effectiveness and efficiency of business decision-making in the agriculture industry? The answer to such a question, or the lack thereof, significantly affects the principal motives of a person working in the agricultural sector, one being achieving substantial profit. The management of diseases, weeds, and irrigation, either in a greenhouse or outdoors, is among the most vital processes a farmer must attend to; AI, alongside the Internet of Things (IoT), may optimize the agricultural practices involved therewith. The IoT consists of a network of physical objects that are embedded with sensors, software, and other technologies for the purpose of connecting and exchanging data with other devices and systems across the internet (Oracle, n.d.). The data processing involved in IoT solutions following the collection and transmission of data by the IoT devices provides a holistic view of the enterprise. AI algorithms are involved to effectuate optimization by multilabel tagging, the reduction of dataset dimensionality, and segmentation (Mueller & Massaron, n.d.). Recommendations relevant to various facets of the work encompassing agriculture may be afforded to a person working in the industry subsequent to the operation of AI algorithms. Concurrently, the utilization of artificially intelligent robotics can automate several of the tasks that must be performed by a farmer and diminish operational costs significantly.
Evidence
Crop, weed, and irrigation management may benefit from a variety of AI implementations relevant thereto. Such applications include the implementation of IoT and robotics with AI algorithms to improve the efficacy and effectiveness of business decision-making in agriculture.
Crop management
A recommender system employing AI algorithms such as SVD and K-Nearest Neighbors can prove efficacious for the management of crops in several ways. Recommender systems use disparate sources of information to provide users with predictions and recommendations of items (Bobadilla et al., 2013). The information utilized by such a system may be processed from data that is collected and transmitted by an IoT infrastructure. Data that would be fed to such a system in the context of agriculture includes water quality and ambient air quality, two critical factors consisting of several parameters for crop management. Figure 1 is an Industrial IoT water pH meter that can monitor pH levels in soil and water within the context of agriculture, horticulture, and other environmental monitoring scenarios (Violet, n.d.). Figure 2 displays an ambient air quality IoT sensor called Polludron, sold by a company named Oizom, that can measure parameters such as temperature, humidity, ambient light, UV radiation, and others (Oizom, n.d.).
The data collected and transmitted by both devices and processed by AI may facilitate pest control and the mitigation of crop diseases in an agricultural scenario. The application of pesticides is regulated in several countries, limiting the field areas that may intake the chemicals. Areas that may not have pesticides applied are called buffer zones and serve to protect surrounding ecosystems from chemical inputs. A machine learning model was developed to identify buffer zones in high-resolution satellite images. Sensitive area data sets, aggregated with regulatory data, pesticide product information, farm machinery information, and other pieces of information, were utilized to produce digital application maps with the buffer zones integrated therein. Given that crop diseases vary by field due to weather conditions and other factors, relevant data is collected and transmitted by IoT to be processed by AI using an expert model to assess the risk of disease for a field. The AI recommender system suggests a period during which pesticides ought to be sprayed, thus managing crop disease. The system also recommends pesticide dosage with field zone pesticide application maps (Janssen et al., 2021). The implementation of AI and delivery of information in such a way renders increased efficiency to a farmer and could result in diminished operating costs.
Weed management
Dwight D. Ligenfelter of the Department of Agronomy at Penn State University defined weeds as plants whose undesirable qualities, according to man, outweigh the good qualities. Weeds are known to diminish crop yields by absorbing water, light, soil nutrients, and space that surrounding plants need; serve as hosts for crop diseases; provide shelter for insects to overwinter; and interfere with harvesting, among other troublesome characteristics and consequences (2009). In an effort to improve the decision-making involved with the application of herbicides to mitigate the effects that weeds have on crop yield, convolutional and multi-layer neural networks, besides cloud computing and unmanned aerial vehicles (UAVs), are utilized to produce a map with the position of weeds on a field. The map is loaded into automated pesticide and herbicide sprayer machines to apply the suggested dosage of chemicals at the specified location (Janssen et al., 2021).
The neural networks applied are particularly useful for image and vision recognition, thus the capability to recognize a weed among other objects in a field. Algorithms such as K-Nearest neighbors and SVD, among others, may be employed within the system to provide chemical dosage and product recommendations. Neural networks, namely one of many AI algorithms, imitate the way biological neurons in the human brain signal to one another. Each node within the network of nodes is a linear regression model composed of input, data, weights, a bias or threshold, and an output. A neural network can improve the accuracy of its output over time, provided it is trained with reliable and sound data. Figure 3 is a depiction of a deep neural network; a neural network is considered deep when it involves more than three layers, including the input and output layer (IBM Cloud Education, 2020).
The implementation of AI in such a manner may result in reduced operational costs due to fewer chemicals being utilized on a field. Furthermore, manual labor hours regarding the management of weeds are diminished as the surveillance, detection, and removal of weeds, along with herbicide dosage and product recommendations are delegated to a machine. Albeit many of the algorithms being implemented besides hardware to remove weeds from a field and commercial application of such a system are still under construction, published research indicates results leading to commercial use. Asselt et al. are involved in the ongoing development of an autonomous weeding robot that works on a sugar beet field to remove weeds using a machine vision algorithm (2006). As the technology progresses, such robots may be placed in a variety of crop fields to remove weeds and not solely a sugar beet field.
Irrigation management
Prudently managing the water God grants to humanity is of utmost importance given the world population, the increase thereof, and the food required to feed everyone, either human or animal. From the freshwater resources currently available, 85% is consumed by the agriculture sector across the world (Patel et al., 2020). AI algorithms to assist with the decision-making involved in irrigation, together with technologies such as IoT and other hardware, collectively afford the irrigation process augmented efficiency, deduced operational costs, and effectiveness to yield crops. Estimating evapotranspiration is recommended to efficiently manage water irrigation and the planning involved therewith. Multiple Artificial Neural Networks (ANNs), each consisting of a feed-forward backpropagation algorithm, have been developed to improve the accuracy of such estimations. Inputs for the algorithm to predict evapotranspiration include temperature, relative humidity, average wind speed, sunshine hours, and rainfall data from Forest Research Institute in Dehradun, India which was collected monthly for 34 years (Chandniha et al., 2017). Withal, researchers from De La Salle University in Manila, Philippines implemented a feed-forward backpropagation Artificial Neural Network (ANN) model consisting of 2 nodes in the input layer, 20 nodes in the hidden layer, and 1 node in the output layer as part of a smart farm automated irrigation system to efficiently allocate water resources (Baldovina et al., 2017).
Drip irrigation is a method of applying water to plants at the root level rather than at the level of the leaves, stems, and fruits. Using water sprinklers is one method to apply water to plants at the level of the leaves, stems, and fruits. The sprinkler method is approximately 65% — 75% efficient at allowing plants to use the water applied, while drip irrigation is 90% efficient therein (University of Rhode Island, 2021). Furman et al. developed Neuro-Drip, an ANN with a statistical description of the spatial-temporal distribution of the water applied by a single drip emitting in a drip irrigation system. The ANN provides illustrations approximating the spatial and temporal subsurface wetting patterns caused by the drip irrigation system (2010). Consequently, the decision-making involved with the application of water to plants may be immensely optimized by the use of a drip irrigation system that has an embedded AI feature.
Analysis
The aforementioned evidence supports the problem being addressed, namely discussing the methods by which AI can improve the effectiveness and efficiency of business decision-making in the agriculture industry. There are three alternative solutions to aid in such an endeavor, including a recommender system that recommends crops to grow in a particular part of a field, an autonomous weeding robot that removes weeds surrounding any type of crop, and an autonomous sowing robot that navigates a plot of land and sows seed in the stead of a person. Table 1 summarizes the possible solutions with corresponding pros and cons. Each alternative solution is characterized by similar pros and cons, with a few exceptions, given that hardware and AI software are involved with all three. From the proposed solutions, one is selected as the optimal and discussed in the final section of the report; the decision is based on what seems to possess superior efficiency and efficaciousness in business decision-making. Among the alternatives.
Table 1: Summary of Alternative Solutions
Crop recommender system
The choice of crop to sow on a sector of land may be exceedingly optimized by the development and deployment of a crop recommender system. Basic data mining techniques such as clustering and classification are executed by machine learning algorithms to effectuate that end, such as K-means, SVM, PCA, and others (Ali & Ali, 2018); K-means and PCA utilize unsupervised learning models and SVM, namely Support Vector Machines, involve supervised learning on regression and classification (Mueller & Massaron, n.d.). There are a variety of such implementations that may reduce operating costs, increase crop yields, and assist in the efficient use of resources. Hepting et al. proffer a web-based decision support system that includes the consideration of climate adaptation to the crop selection process. The solution allows for crop recommendations that are highly correlated to a specific piece of land and corresponding characteristics which are analyzed by data that is collected with an on-premises or nearby sensor network (2012). The system proposes a crop recommendation to someone working in the context of agriculture that is suited to a specific location when utilizing the on-premises option for the sensor network, as opposed to using data collected from a disparate and nearby location that may not be as accurate. In either case, both types of sensor networks allow for improved decision-making in the selection of crops to sow.
Data from the Bangladesh Agricultural Research Institute (BARI) was availed of by Ahamed et al. to predict crop yield in Bangladesh. The researchers selected five crops and used artificial intelligence composed of multiple linear regression, k-nearest neighbor, k-means, and artificial neural network machine learning algorithms. Four clusters were initialized by the k-means algorithm; cluster one was based on weather attributes, such as the amount of rainfall, minimum temperature, maximum temperature, humidity, and degree of sunshine; cluster two was based on soil pH and soil salinity; cluster three was based on areas that were irrigated; and cluster four was based on individual crop yields of the five that were selected for the project, namely aman, aus, boro, potato, and wheat. Given the dataset that was used by the researchers, it was discovered that artificial neural networks provide better predictions for wheat, potato, and aus, while multiple linear regression performed better in predicting boro and aman; the root means square error (RMSE) statistic was used to determine the suitableness of each model (2015). The process outlined holds a vital role in the implementation of artificial intelligence within a crop recommendation system to improve the decision-making involved with crop selection. Yet, there was another project conducted by researchers at the National Institute of Technology (NIT) located in India. The project involved the use of the same algorithms for the project in Bangladesh with the addition of a support vector machine (SVM), random forest, and gradient boosting. The researchers presented a technique called Crop Selection Method (CSM) to recommend a sequence of crops to plant throughout a season. The input data for the CSM algorithm was gathered from farmers located in Patna district at Bihar, India (Kumar et al., 2015).
Albeit the three systems avail collected data fairly well to optimize the selection of crops to sow, the time required to develop such a system may impose a large sum of money or other resources on an individual or group investing in such a technology. Ongoing system maintenance costs which may include updates and other work are necessary, further increasing the cost to maintain and use the system.
Autonomous weeding robot
As previously mentioned, the presence of weeds on a field may drastically mar the yield of a given crop due to vital resources being consumed by the surrounding weeds. The incorporation of an autonomous weeding robot can mitigate the harm caused by weeds through its operation in removing the undesired plant. Given the technology is still in development, agriculture personnel has invested much time and money to apply pesticides, thus putting a stop to the damage caused by weeds. Unfortunately, the chemicals applied to weeds may adversely affect nearby plants; a phenomenon called herbicide damage. Herbicide damage refers to the adverse and undesired effects a plant exhibits that are caused by the application of pesticide thereto designed for weed control, such as an herbicide (Hudelson, 2010). An autonomous weeding robot may significantly reduce such damage as the need to apply the chemicals to a plant for the purpose of extinguishing weeds is removed. Moreover, removing weeds involves manual labor that requires time and resources. The automated removal of weeds should optimize the process and diminish the hours required to manage weeds while preserving resources. Several researchers have undertaken the development of autonomous weeding robots with successful results.
Chaihan et al. reported on the construction and implementation of an autonomous weeding robot that was developed at the Asian Institute of Technology (AIT) in collaboration with Chulalongkorn University in Thailand and the Agricultural Research and Development Association (ARDA) of Thailand. The robot was built to manage the weeds on a rice production field in Thailand. The physical structure of the robot included an autonomous navigation algorithm that was developed by the researchers; it was complemented by a global navigation satellite system (GNSS) and a compass. The robot performed well in ideal circumstances but exhibited diminished performance when met with high weed density. Chaihan et al. note that the main disadvantage of the system was a low-cost GNSS sensor which caused the robot to skew from its path (2019). Moreover, an autonomous weeding robot project conducted by Asselt et al. was aforementioned; the robot was designed and developed to be implemented on a sugar beet field (2006). Other academics have engaged in the development of an autonomous weeding robot but will not be mentioned for the sake of brevity. Figure 4 displays the conceptual model of the weeding robot designed at AIT, while Figure 5 presents the autonomous weeding robot developed to operate on a sugar beet field.
The autonomous machines present the potential for decreased operating costs, manual labor hours, harm to crops, and increased crop yields, but the high implementation cost may discourage farmers to implement the device in production. Provided that research and development continue, perhaps the present errors in the systems may be removed, but the production and sale of such robots are currently hindered thereby. Maintenance on the robots seems irrelevant at the present stage of development yet present a potential downside to the adoption of such machines in agriculture if the maintenance required is expensive and arduous. Akin to the weeding robot is the development of an autonomous sowing robot.
Autonomous sowing robot
Sowing seeds on a field is characterized by labor-intensive work that may be optimized with the help of an autonomous robot performing the task. Financial and other resources may be preserved due to fewer expenditures thereof on manual labor to sow the desired seeds. Precision agriculture is the typical goal of such robots, namely a management strategy that gathers, processes, and analyzes temporal, spatial, and individual data and combines it with other information to support management decisions (The International Society of Precision Agriculture, 2019); the concept of precision agriculture may be applied to the previously discussed technologies in the crop and weed management sections of the report. Researchers have participated in research and development of autonomous sowing robots with successful developments thereof, albeit further work is required to introduce a production-grade product to the agriculture market. The robot designed and developed by Gladwin et al. is an example whereby a mobile robot capable of sowing seeds autonomously on a piece of land was fabricated. Figure 6 is a picture of the sowing robot constructed by Gladwin et al. The structure of the robot provides easy maintenance of the robot, but the accuracy of its traversal on a piece of land was lacking at the time the project was reported and published (2017). A sowing robot constructed by Chenchela et al. is another example whereby the sowing process is optimized and automatized; the robot can navigate on any agricultural land and perform seed sowing simultaneously. Hardware such as onboard sensors, cameras, a personal computer, and a suspension system, along with the integration of a global position system (GPS) and multiple algorithms are utilized by the robot to perform the necessary functions required to navigate a piece of land and sow seed (2018).
As with the autonomous weeding robot, an array of benefits may be availed of from the use of such technology, but the cost associated with research, development, implementation, and maintenance of a production-grade sowing robot may discourage farmers from incorporating the machine into production. Furthermore, the sowing robots presented by the research groups mentioned are not production-ready, thus disallowing the use of benefits proposed therefrom.
Conclusion
Recognizing the technological advancements involved with each subsystem in the AI implementations previously discussed, a solution is proposed wherein the functionality of the autonomous sowing robot and the autonomous weeding robot are combined to have a robot where the weeding and sowing agricultural processes are handled in one system. The research, development, and implementation of such a system that combines the weeding and sowing function aggregates the relevant processes to one system, thus reducing any associated maintenance costs and the use of resources to one system as opposed to two disparate systems. Environmental data collected from an IoT infrastructure and other remote devices may be fed to the autonomous robot, further assisting the sowing function by recommending appropriate seeds and executing the planting function accordingly.
Discovering ways by which the efficiency and effectiveness of decision-making involved in agriculture may be improved by AI was the scope of the report. Agricultural business processes such as crop, weed, and irrigation management are significantly affected consequent to the implementation of AI thereto; the lack of such implementations implies the continuance of inefficiencies and ineffectiveness wrought by an absence of AI in the context of agriculture. Evidence of potential contemporary applications of AI for crop, weed, and irrigation management were proffered, such as algorithms utilizing data collected from an internet-connected pH meter and air quality monitor to recommend pesticide dosage for a particular part of a field, among other examples presented for crop, weed, and irrigation management. Considering the evidence for the use of AI to improve decision-making in agriculture, a crop recommender system, autonomous weeding robot, and autonomous sowing robot were suggested as alternative solutions to the problem that was addressed. As research and development are ensued, the summation of subsystems to construct a system wherein the sowing and weeding functionality are executed by one system may materialize as a production-ready device and cause the optimization of agricultural work.
References
1. Ahamed, A. T. M. S., Mahmood, N. T., Hossain, N., Kabir, M. T., Das, K., Rahman, F., & Rahman, R. M. (2015, June 1). Applying data mining techniques to predict annual yield of major crops and recommend planting different crops in different districts in Bangladesh. 2015 IEEE/ACIS 16th International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD), 1–6. https://doi.org/10.1109/SNPD.2015.7176185
2. Ali, S. M., & Ali, M. (2018, February). A Survey on Agriculture Crop Recommender System. International Journal of Advanced Research in Computer Engineering & Technology (IJARCET), 7(2), 218–220.
3. Asselt, K. V., Bakker, T., Bontsema, J., Miller, J., & Straten, G. V. (2006). An Autonomous Weeding Robot for Organic Farming. Springer Tracts in Advanced Robotics, 25. https://doi.org/10.1007/978-3-540-33453-8_48
4. Baldovino, R. G., Bandala, A. A., Dela Cruz, J. R., & Dadios, E. P. (2017, October 19). Water usage optimization of Smart Farm Automated Irrigation System using artificial neural network. 2017 5th International Conference on Information and Communication Technology (ICoIC7), 1–5. https://doi.org/10.1109/ICoICT.2017.8074668
5. Bobadilla, J., Ortega, F., Hernando, A., & Gutiérrez, A. (2013, April 6). Recommender systems survey. Knowledge-Based Systems, 46, 109–132. https://doi.org/10.1016/j.knosys.2013.03.012
6. Chaihan, R., Ekpanyapong, M., & Kanagasingham, S. (2019, November 16). Integrating machine vision-based row guidance with GPS and compass-based routing to achieve autonomous navigation for a rice field weeding robot. Precision Agriculture 21, 831–855. https://doi.org/10.1007/s11119-019-09697-z
7. Chandniha, S. K., Khare, D., & Nema, M. K. (2017, March 2). Application of artificial intelligence to estimate the reference evapotranspiration in sub-humid Doon valley. Applied Water Science, 7, 3903–3910. https://doi.org/10.1007/s13201-017-0543-3
8. Chenchela, K. R. V., Kapileswar, N., Santhi, V. P., & Prasad, S. V. C. H. (2018, June 21). Sensor and vision-based autonomous AGRIBOT for sowing seeds. 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS), 242–245. https://doi.org/10.1109/ICECDS.2017.8389873
9. Furman, A., Hinnell, A. C., Lazarovitch, N., Poulton, M. & Warrick, A. W. (2010, March 9). Neuro-Drip: estimation of subsurface wetting patterns for drip irrigation using neural networks. Irrigation Science, 28, 535–544. https://doi.org/10.1007/s00271-010-0214-8
10. Gladwin, J. S., Natarajan, R. A., Prabhu, P., Rajavel, R., Sivaraman, V. N., Smruthi, S. S., & Srinivasan, N. (2017, April 24). Design of an autonomous seed planting robot. 2016 IEEE Region 10 Humanitarian Technology Conference (R10-HTC), 1–4. https://doi.org/10.1109/R10-HTC.2016.7906789
11. Goehring, B. & Marshall, A. (2020, November). The business value of AI. IBM. https://www.ibm.com/downloads/cas/ZENVBND4
12. Hepting, D. H., Hill, H., & Maciag, T. (2012, February 9). Web-Based Support of Crop Selection for Climate Adaptation. 2012 45th Hawaii International Conference on System Sciences, 1227–1236. https://doi.org/10.1109/HICSS.2012.640
13. Hudelson, B. (2010, May 11). Herbicide Damage. Wisconsin Horticulture — Division of Extension. https://hort.extension.wisc.edu/articles/herbicide-damage/
14. IBM Cloud Education. (2020, August 17). Neural Networks. IBM. https://www.ibm.com/cloud/learn/neural-networks
15. Janssen, O., Selinger, S., Shankar, P., & Werner, N. (2021, January 12). Artificial Intelligence Driven Crop Protection Optimization for Sustainable Agriculture. 2020 IEEE / ITU International Conference on Artificial Intelligence for Good (AI4G), 1–6. https://doi.org/10.1109/AI4G50087.2020.9311082
16. Kumar, P., Kumar, R. Singh, J. P., & Singh M. P. (2015, August 27). 2015 International Conference on Smart Technologies and Management for Computing. Communication, Controls, Energy, and Materials (ICSTM), 138–145. https://doi.org/10.1109/ICSTM.2015.7225403
17. Ligenfelter, D. D. (2009, December 9). Introduction to Weeds: What are Weeds and Why do we Care? PennState Extension. https://extension.psu.edu/introduction-to-weeds-what-are-weeds-and-why-do-we-care
18. McCarthy, J. (n.d.). What is AI? / Basic Questions. Stanford. http://jmc.stanford.edu/artificial-intelligence/what-is-ai/index.html
19. Mueller, J. P. & Massaron, L. (n.d.). Machine Learning For Dummies Cheat Sheet. Dummies. https://www.dummies.com/programming/big-data/data-science/machine-learning-dummies-cheat-sheet/
20. Oizom. (n.d.). Ambient Air Quality Monitoring System. Oizom. https://oizom.com/product/polludrone-air-pollution-monitoring/
21. Oracle. (n.d.). What is IoT? Oracle. https://www.oracle.com/internet-of-things/what-is-iot/
22. Patel, N., Shah, D., Shah, M., & Talaviya, T. (2020, April 22). Implementation of artificial intelligence in agriculture for optimization of irrigation and application of pesticides and herbicides. Artificial Intelligence in Agriculture, 4, 58–73. https://doi.org/10.1016/j.aiia.2020.04.002
23. The International Society of Precision Agriculture. (2019). Precision Agriculture. Springer. https://www.springer.com/journal/11119
24. University of Rhode Island. (2021). Drip Irrigation. University of Rhode Island. https://web.uri.edu/safewater/protecting-water-quality-at-home/sustainable-landscaping/drip-irrigation/
25. Violet. (n.d.). New Product Post: Industrial pH Sensor, Soil Moisture Sensor, EC Sensor and More. Seeed Studio. https://www.seeedstudio.com/blog/2020/06/19/new-product-post-industrial-ph-sensor-soil-moisture-sensor-ec-sensor-and-more/
