During my senior year of undergraduate studies, I designed a system for remotely monitoring crop farming utilizing Internet of Things (IoT) technology and cloud services. This innovation aims to assist farmers in implementing precision farming practices.
Project Description
Tools Used:
Problem Statement
The current state of large-scale farming in many areas in Kenya is characterized by outdated and unreliable methods, leading to significant challenges for farmers who lack essential knowledge. Predictive models often fail, resulting in substantial financial losses for the farmer. Crucial factors like soil moisture, air quality, and precise irrigation are not systematically addressed, with many farmers relying on intuition rather than adopting precision farming techniques.
In Kenya, existing agricultural systems lack remote monitoring capabilities, and advanced technologies are typically designed for specific organizations, excluding local farmers. To address these issues, our proposed solution leverages IoT to introduce a new standard of reliability and accessibility in crop monitoring and intelligent farming for the agribusiness sector.
Objectives
To Formulate requirement analysis specifications for remote monitoring systems using sensors and Arduino.
To design a system that after collecting data, stores it on a dashboard and displays real-time data on a dashboard
To build a smart monitoring system using Arduino and cloud services.
To implement a user-friendly dashboard where the farmer can quickly start the sensors and view the data in real-time
To deploy the system on a model and demonstrate its full functionality.
Observation
Interviews
Desktop Research
Surveys
Persona
Information Architecture
Research Methods
In my research, I conducted a comprehensive desktop review, analyzing prior academic works and examining existing solutions offered by prominent companies like Microsoft (Farmbeats) and Liquid Telecom. To gain deeper insights, I interviewed a key developer from the Kenya Agricultural and Livestock Research Organisation. Using a snowball sampling technique, I expanded my interview, observation and survey pool, ultimately reaching the desired sample size and scope. The synthesized data was then organized into thematic groups, allowing for the development of relevant personas and user journey maps. These insights played a crucial role in informing the proposed solution and shaping the "how might we" statement, providing a robust foundation for addressing the identified agricultural challenges.
Solution
The proposed solution suggests having two stations for the farmer's convenience. One station would be located locally on the farm, allowing the farmer to physically manipulate it. This local station would provide a hands-on approach for the farmer to control and manage various tasks related to the farm. On the other hand, the second station would be a remote station accessible through the web or a phone application. This remote station would enable the farmer to manipulate and control farm operations from a distance, providing convenience and flexibility. With this setup, the farmer could efficiently manage the farm both on-site and remotely, ensuring optimal productivity and convenience.
Local vs. Remote Station
Design Methods
The proposed solution led to a comprehensive requirements analysis for the system, followed by the creation of system design elements such as use case models, flow of events, sequence flow models, and more. These steps were taken to provide clear guidance for the systematic development of the system. Iteration is key during this process.
Outcomes
The project implemented a comprehensive remote monitoring system utilizing ESP8266 Wi-Fi Module that uses NodeMCU as the central controller, seamlessly interfacing with various sensors including the DHT11 for temperature and humidity and a soil moisture sensor. ESP8266 Wi-Fi Module efficiently communicated with these sensors, collecting real-time data that was transmitted to the ThingSpeak IoT platform via Wi-Fi connectivity. ThingSpeak served as a centralized hub for storing and managing and visualizing the collected data. A dedicated web page was designed to dynamically presented the sensor data from ThingSpeak, offering intuitive web control from the user. Similarly, did the Blynk App offer the same internet control to the user. The integration of these tools demonstrated a cohesive system where ESP8266 Wi-Fi Module orchestrated data flow, ThingSpeak acted as a repository, and the web page and Blank App provided a user-friendly interface for informed decision-making in plant care, showcasing the reliability and effectiveness of the IoT-based solution.
Learnings
The project successfully developed a remote plant monitoring system using the NodeMCU IoT platform, integrating various sensors, data transmission to ThingSpeak, and a dedicated web page for real-time monitoring of temperature, humidity, and soil moisture levels. The NodeMCU proved to be a reliable and cost-effective solution for data processing and communication. The project demonstrated effective remote plant monitoring, emphasizing informed decision-making in plant care through secure cloud-based data storage and user-friendly data visualization. Testing validated the functionality and interaction of components, showcasing achievements in data visualization, ThingSpeak integration, and open-source technology utilization. The project contributes to IoT-based plant monitoring, suggesting opportunities for further advancement.
Recommendations
Looking forward, it would be beneficial incorporating machine learning or AI techniques for enhanced capabilities and intelligent plant care. Predictive analysis using time series forecasting algorithms like ARIMA or LSTM can forecast future conditions, optimizing watering schedules. Reinforcement learning algorithms can learn optimal watering schedules, and anomaly detection algorithms can identify irregular sensor patterns, prompting timely interventions. Personalized plant care recommendation systems can use machine learning to tailor advice based on user preferences, plant species, and sensor data. These recommendations reflect a pathway for future work, emphasizing the potential of machine learning to further improve the efficiency and effectiveness of remote plant monitoring and smart agriculture applications.