Decentralized irrigation system, device and mehtod using esp-now for multi-crop, zone-based, offline agricultural water management

Abstract

ABSTRACT DECENTRALIZED IRRIGATION SYSTEM, DEVICE AND MEHTOD USING ESP-NOW FOR MULTI-CROP, ZONE-BASED, OFFLINE AGRICULTURAL WATER MANAGEMENT The present invention discloses an adaptive and autonomous multi-zone irrigation system (500) for precision water management in agricultural fields. The system comprises a plurality of ESP32 slave devices (502-1 to 502-N), each associated with a crop zone (504-1 to 504-N) and embedded with sensors (506-1 to 506-5) including soil moisture, temperature-humidity, EC, pH, and image sensors. A first microcontroller unit (508-1 to 508-N) transmits sensor data to a master ESP32 device (510) via ESP-NOW protocol (512). The master device includes a second microcontroller unit (514) that uses on-device machine learning and machine vision to process real-time and historical data to determine irrigation needs. Based on analysis, an irrigation control signal actuates dynamic valve assemblies (516-1 to 516-N) comprising gear-driven three-way valves (518-1 to 518-N) and controllers (520-1 to 520-N) using PID or AI-based feedback. This system enables water source selection, soil-health correlation, and efficient, zone-specific irrigation control. FIG. 5 will be the reference figures.

Description

Description:FORM 2THE PATENT ACT, 1970(39 of 1970)&THE PATENT RULES, 2003COMPLETE SPECIFICATION(See section 10; rule 13)“DECENTRALIZED IRRIGATION SYSTEM, DEVICE AND MEHTOD USING ESP-NOW FOR MULTI-CROP, ZONE-BASED, OFFLINE AGRICULTURAL WATER MANAGEMENT”Rohan Shamkant Chavan an Indian citizen, having address of Atharva Apartment, Sector 1, Indrayaninagar, Bhosari, Pune – 411039Prasad Vinod Ahirrao an Indian citizen, having address of Prathamesh Vihar C-203, Anganwadi Chowk, Krushnanagar, Chinchwad Pune-411019Rohit Ravikiran Digole an Indian citizen, having address of 958, At Po. Rohina, Tq. Chakur, Latur-413517Komal Sunil Aujekar an Indian citizen, having address of New krushanarpan colony, Amravati-444605Kaushal Chandrashekhar Lawande an Indian citizen, having address of I-703, River Residency, Chikhali, Pune-411062THE FOLLOWING SPECIFICATION PARTICULARLY DESCRIBES THE INVENTION AND THE MANNER IN WHICH IT IS TO BE PERFORMEDTECHNICAL FIELD

[0001] The present invention relates to the field of precision agriculture and smart irrigation systems. More specifically, it pertains to an automated, zone-based irrigation management system for multi-crop farming using ESP-NOW-based local communication and decentralized sensor architecture.BACKGROUND

[0002] Agriculture remains the foundation of economic stability in several countries, particularly in India, where a substantial portion of the population is engaged in farming activities. Despite the gradual incorporation of technology in various sectors, irrigation practices continue to rely heavily on traditional and labor-intensive methods. Common practices like flood irrigation, manual watering, and unregulated sprinkler use often result in overwatering, uneven water distribution, high water wastage, and significant manual labor dependency.

[0003] Several smart irrigation systems have been developed to overcome the inefficiencies of traditional methods. Typically, these systems integrate soil moisture, temperature, and humidity sensors with microcontrollers (such as NodeMCU) and cloud-based control platforms.

[0004] While such systems improve remote monitoring and water-use efficiency, they come with inherent limitations that hinder wide-scale adoption in practical farming environments. Firstly, these systems are heavily reliant on stable internet connectivity and cloud platforms for core operations and decision-making. However, reliable internet infrastructure is often lacking in rural or remote agricultural regions, rendering such systems ineffective when connectivity fails. Secondly, many existing smart irrigation systems are designed for uniform crop fields and cannot dynamically adjust water distribution based on zone-specific crop needs. This limitation renders them unsuitable for multi-crop environments, where precise water control is critical. Additionally, subscription-based cloud services, proprietary components, and GSM modules significantly increase setup and maintenance costs—making these solutions unaffordable for small and marginal farmers.

[0005] Furthermore, most current systems require continuous power supply and involve energy-intensive operations, which pose additional challenges in off-grid locations. Communication protocols like LoRa, Bluetooth, or Wi-Fi often require dedicated infrastructure and add to system complexity. These cumulative issues underline the absence of a robust, low-cost, offline-compatible, and crop-sensitive irrigation solution in the current landscape.

[0006] These limitations become more pronounced in fields where multiple crops with differing water needs are cultivated simultaneously. In such scenarios, manual or uniform irrigation systems frequently lead to suboptimal watering—over-irrigating some crops while under-irrigating others. This imbalance not only hampers plant health but also results in reduced agricultural yield and inefficient use of water resources. To date, no widely adopted solution adequately addresses the irrigation complexities associated with diverse, multi-crop cultivation.

[0007] In view of the foregoing challenges, there exists a pressing need for a cost-effective, energy-efficient, and reliable smart irrigation system that supports real-time, zone-specific irrigation decisions, functions without continuous internet or cloud reliance, facilitates local communication among sensor nodes and control units, accommodates multi-crop irrigation management using minimal infrastructure, remains operable in off-grid and remote locations with inconsistent power supply, and offers low maintenance and user-friendly control for non-technical farmers.SUMMARY

[0008] The present invention fulfills these needs by providing an offline-capable, peer-to-peer, solar-powered irrigation management platform built on ESP32 microcontrollers. It introduces a decentralized, zone-based architecture capable of autonomously managing water flow to different crop zones based on real-time soil moisture readings.

[0009] The invention comprises a master-slave configuration of ESP32-based microcontroller units. Each slave node is strategically placed within distinct crop zones and is equipped with soil moisture, temperature, and humidity sensors. These nodes operate independently, collecting environmental data specific to their assigned zones and transmitting it to a central Master Node using the ESP-NOW communication protocol—an energy-efficient, peer-to-peer wireless protocol that operates without Wi-Fi routers or cloud servers.

[0010] The Master Node aggregates the data, processes it locally, and initiates irrigation for only those zones where the soil moisture falls below a predefined crop-specific threshold. Actuation is managed through a relay-controlled water pump and a low RPM DC gear motor coupled with a three-way valve, which dynamically channels water to the appropriate crop zones. This enables fine-grained water distribution based on individual crop requirements—something most existing systems fail to address.

[0011] To enable offline autonomy, all components are powered using solar panels with integrated lithium-ion batteries and Battery Management Systems (BMS). The ESP32 modules utilize deep sleep modes to minimize power consumption and extend operational life in off-grid environments. Optional Wi-Fi connectivity allows for data upload and remote access when available, but it is not required for core functionality.

[0012] Unlike conventional irrigation systems that rely on timer-based or manual control mechanisms, the disclosed invention offers real-time, adaptive irrigation using a network of wireless sensor devices and AI-driven control. The integration of machine learning on a distributed microcontroller platform, capable of operating in low-power deep sleep mode, represents a unique technical architecture not seen in prior art.

[0013] The system's capability to combine multi-modal sensor inputs with image-based machine vision analytics is the advancement of the present invention. This allows the system not just to measure soil or atmospheric conditions, but to interpret biological signals such as leaf wilting or stress—something far beyond conventional soil moisture sensors or remote monitoring systems. The selection of irrigation source based on EC and pH, and dynamic valve modulation using PID or AI feedback, further enhances system intelligence.

[0014] From a technical advancement standpoint, the system provides an edge over traditional or semi-automated irrigation systems through energy-efficient decentralized data collection, intelligent decision-making using on-device models, and fine-grained, zone-specific control of water distribution. This ensures not just operational automation but optimization and sustainability, making it highly aligned with the goals of smart agriculture and precision farming.

[0015] In terms of non-obviousness, a person skilled in the art would not arrive at this solution without inventive ingenuity. The integration of machine learning and machine vision into a microcontroller-based system operating on limited power and transmitting via ESP-NOW protocol is non-trivial. Furthermore, the decision-making loop—starting from data capture to actuation based on learned models—demonstrates a systemic innovation rather than a mere aggregation of known components.

[0016] Technical benefits include: precision irrigation with reduced water usage; improved crop health monitoring and yield; energy efficiency through deep-sleep enabled esp32 devices; soil-specific irrigation logic based on real-time and historical data correlation; and intelligent water source selection improving resource optimization.

[0017] This invention provides a practical, scalable, and farmer-friendly alternative to conventional smart irrigation systems. It significantly advances the field of precision agriculture by making intelligent irrigation accessible to small and marginal farmers, especially those growing multiple crops with diverse water requirements.BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGS

[0018] FIG. 1 illustrates exemplary block diagram of smart irrigation system, in accordance with an embodiment of the present disclosure.

[0019] FIG. 2 illustrates exemplary flow chart of smart irrigation system, in accordance with an embodiment of the present disclosure.

[0020] FIGs. 3A-3D illustrates an exemplary experimental set up for the implementation, showing user interface, exemplary crop zone, and mechanical connection between the components, in accordance with an embodiment of the present disclosure.

[0021] FIG. 4 illustrates an exemplary computer system to implement functionalities of the system for generating applications in real-time based on multimodal inputs, in accordance with embodiments of the present disclosure.

[0022] FIG. 5 illustrates an exemplary block diagram showing irrigation system for adaptive and autonomous multi-zone water management in agricultural fields, in accordance with embodiments of the present disclosure.

[0023] FIG. 6 illustrates an exemplary method for adaptive and autonomous multi-zone water management in agricultural fields, in accordance with embodiments of the present disclosure.DETAILED DESCRIPTION

[0024] The invention pertains to an automated solution designed to optimize water usage and enhance agricultural efficiency. It utilizes the ESP-NOW protocol for wireless communication between multiple ESP32 modules arranged in a master-slave configuration. This allows for real-time monitoring of soil moisture, temperature, and humidity. The system operates without internet connectivity for local communication, while also supporting cloud-based remote monitoring when internet access is available. By automating irrigation control, system reduces water waste, improves crop health, and decreases the need for manual intervention. Many farmers with small fields (in acres) grow multiple crops, each requiring different irrigation needs. Traditional manual methods require significant labor and precise scheduling, often leading to inefficient water use and adversely affecting crop health. Moreover, existing smart irrigation solutions can be expensive and heavily rely on internet connectivity, making them impractical for remote areas. The invention addresses these challenges with a low-cost, solar-powered system. This ensures efficient water management and supports sustainable farming practices, making it an ideal solution for modern agriculture.

[0025] The invention pertains to an advanced, adaptive, and autonomous irrigation system specifically designed for efficient water management across multiple crop zones in agricultural fields. The system is built around a decentralized architecture that includes a network of slave and master devices based on Espressif 32 (ESP32) microcontrollers. Each slave device is deployed in a corresponding crop zone and is equipped with multiple environmental and soil-specific sensors, including soil moisture, pH, electrical conductivity, temperature, humidity, and an image sensor. These devices periodically wake from deep sleep to collect and transmit data using ESP-NOW protocol, enabling energy-efficient wireless communication.

[0026] At the core of the system is a master device that aggregates real-time data from all the slave nodes. It uses an embedded machine learning model trained on historical and current sensor data to determine the precise irrigation needs of each zone. Additionally, a machine vision model processes the image data to evaluate crop canopy features such as stress indicators and leaf area index. The system intelligently correlates visual data with environmental metrics to generate control signals that modulate zone-specific water delivery using gear-driven, PID- or AI-controlled dynamic valves. The other components include water intake modules capable of selecting optimal water sources based on quality parameters and a pump for water delivery. This multi-sensor, machine learning-enabled system ensures precise, need-based irrigation to enhance crop yield and conserve water resources.

[0027] In operation, each slave device is installed within a defined crop zone. The slave device remains in a deep sleep state to conserve power and periodically activates to gather data from its attached sensors. These include sensors for soil moisture, pH, temperature, humidity, electrical conductivity, and crop images. Upon activation, the slave device collects sensor data and transmits it wirelessly to the master device using the ESP-NOW protocol, a low-power, peer-to-peer wireless communication standard optimized for ESP32 chips.

[0028] The master device receives sensor and image data from all active slave devices. It runs an on-board machine learning model that analyzes environmental trends and predicts soil moisture depletion based on time-series data. Concurrently, it processes image data using a vision model that identifies plant health indicators such as wilting or stress through features like leaf area index or NDVI. This data fusion allows the system to make a well-informed decision about whether and how much to irrigate each zone.

[0029] Based on the inference, the master device generates control signals. These signals are sent to the respective dynamic valve assemblies in each crop zone. Each valve assembly includes a gear-driven three-way valve that is dynamically modulated using either a PID controller or an AI-based feedback control loop. The actuation of these valves determines the volume and direction of water flow to each zone, tailored to the specific needs of the soil and crop condition in that area.

[0030] In zones where, multiple water sources are available, an intelligent water intake module selects the most suitable water source. The selection logic prioritizes sources based on real-time pH and EC values, ensuring that the chemical properties of water are compatible with the crop's needs. If needed, a pump module is activated to deliver water from the selected source to the zone.

[0031] Advantages of the present invention:

[0032] Fully Offline Capable: Operates without continuous internet or cloud dependence.

[0033] Zone-Based, Multi-Crop Irrigation: Supports heterogeneous crop fields with variable irrigation needs.

[0034] Energy Efficient: Powered by solar energy with deep sleep optimization.

[0035] Low Cost and Scalable: Minimal hardware requirements and no recurring cloud fees.

[0036] Hybrid Control System: Offers both automatic and manual control via user interface.

[0037] Robust in Remote Areas: Designed to function reliably in low-resource, off-grid environments.

[0038] FIG. 1 illustrates exemplary block diagram of smart irrigation system, in accordance with an embodiment of the present disclosure.

[0039] The invention relates to an intelligent, adaptive, and autonomous irrigation system designed for precision water management in agricultural fields. As shown in FIG. 1, the system utilizes a distributed architecture composed of multiple ESP32-based slave nodes and a centralized ESP32 master node. Each slave unit is strategically deployed in a specific crop zone and is equipped with a suite of environmental sensors, including a soil moisture sensor, a temperature sensor, and a humidity sensor. These sensors monitor key agro-environmental parameters in real time. To support operation in remote, off-grid agricultural locations, each slave unit is powered by a solar panel and a rechargeable battery. The ESP32 microcontroller embedded in each slave device is programmed to remain in deep sleep mode to minimize power consumption and is periodically activated at fixed intervals to collect sensor data. The data is then transmitted wirelessly to the master unit using ESP-NOW, a low-power, peer-to-peer communication protocol specifically optimized for ESP32 systems.

[0040] The master unit is centrally located and is also powered by its own solar panel and battery, ensuring self-sufficiency and uninterrupted operation. The master device incorporates a real-time clock for timestamping incoming data and an ESP32 microcontroller for handling communication, decision-making, and actuation control. Upon receiving sensor data from multiple slave nodes, the master processes the inputs locally. It executes embedded algorithms or lightweight machine learning models trained on historical and real-time sensor data to determine the irrigation requirements of each crop zone. The system does not rely on cloud processing for decision-making, thereby reducing latency and ensuring real-time responsiveness even in regions with limited connectivity. In addition to soil and environmental sensing, the system supports image-based monitoring, enabling the detection of crop canopy features such as wilting or stress, further refining irrigation decisions.

[0041] Based on the analysis, the master device generates control signals to manage irrigation hardware. A key component of the actuation mechanism is a low-RPM DC gear motor connected to a three-way irrigation valve. This valve modulates the water flow to individual crop zones depending on the calculated need. The gear motor is driven through an L298 motor driver and controlled via a relay circuit managed by the master’s microcontroller. To ensure precision in valve movement, the system incorporates an encoder that provides feedback on the motor's position. This feedback enables dynamic modulation of valve position through a Proportional-Integral-Derivative (PID) control loop or, optionally, an AI-based feedback mechanism, resulting in highly accurate zone-specific water distribution.

[0042] Further, the master unit is capable of uploading processed data to a cloud server through Wi-Fi or cellular networks. This enables real-time data visualization and remote control through a mobile application or web dashboard. Users can access environmental conditions, irrigation history, and system status remotely and can override the system when necessary. This functionality significantly improves the transparency, manageability, and decision-making capabilities for end users such as farmers, agronomists, or agricultural service providers.

[0043] The system's hardware is optimized for low power operation and long-term deployment. Solar panels continuously recharge the onboard batteries, while intelligent power management using deep sleep and selective wake cycles ensures energy efficiency. The communication protocol, ESP-NOW, allows multiple devices to operate without the need for complex Wi-Fi infrastructure, reducing setup costs and making the system scalable for large farms with diverse irrigation needs. Each component—from sensor input to water delivery—is seamlessly integrated to provide a fully automated, modular, and responsive irrigation solution.

[0044] Overall, the invention as implemented in FIG. 1 demonstrates a technically advanced and logically structured system that embodies the principles of smart agriculture. It integrates real-time environmental sensing, edge-level machine intelligence, low-power communication, and smart actuation into a coherent solution for optimizing water usage in agriculture. By addressing the challenges of labor-intensive manual irrigation and inefficient water use, the system contributes significantly to sustainability, crop health, and yield optimization.

[0045] Referring again to FIG. 1, the system is Divided into following functional modules:

[0046] Master Device (Control Unit):

[0047] The Master device serves as the central component of the smart irrigation system, coordinating operations related to data collection, processing, decision-making, and actuation. Its role involves analyzing environmental data in real time and managing irrigation activities across multiple crop zones, contributing to efficient water usage.

[0048] Data Collection: The Master Node utilizes the ESP-NOW protocol for efficient and low-latency wireless communication with Slave Nodes. It gathers sensor readings after specific period. Data includes soil moisture levels, temperature, and humidity. To enhance logging accuracy, each data packet includes a unique identifier for the sensor node along with timestamps provided by a Real-Time Clock (RTC), ensuring precise tracking of environmental conditions.

[0049] Data processing and decision making: After collecting the data, the Master Node analyzes the soil moisture readings against predefined thresholds tailored to specific crop requirements. When moisture levels fall below the designated threshold for a particular crop zone, the Master Node initiates the irrigation process. This critical decision-making takes place locally on the ESP32 microcontroller, ensuring that the irrigation system remains functional even in the absence of internet connectivity.

[0050] Irrigation Control: Upon determining the need for irrigation, the Master Node activates a relay module connected to the water pump. It manages water distribution through the use of a low RPM DC gear motor connected with a three-way valve. Controlled by a motor driver, along with an encoder, the gear motor allows for precise rotation, ensuring the valve aligns accurately to the targeted crop zone. This dynamic system enables automated irrigation without the need for manual intervention, facilitating efficient water management.

[0051] Data Monitoring (Cloud upload): When internet connectivity is available (Wi-Fi or cellular networks), the ESP32 establishes a connection to a cloud platform. The system uploads real-time sensor data, irrigation status, and timestamps, making this information accessible through a user-friendly webpage or mobile application. Users can easily view environmental data, monitor irrigation status and historical logs, manually override the system if necessary, and adjust irrigation thresholds or schedules remotely.

[0052] Power Management: The Master Node operates on a solar panel and rechargeable li-ion battery along with BMS, which enables continuous functionality in off-grid agricultural environments.

[0053] Slave Nodes (Field Nodes):

[0054] The Slave device is a key part of the smart irrigation system, responsible for collecting environmental data from different crop zones. It measures soil moisture, temperature, and humidity, then sends this data to the Master device using ESP-NOW. This helps the system make accurate decisions for zone-based irrigation.

[0055] The Slave Nodes are placed in various crop zones to gather local environmental data and transmit it to the Master Node. Each node operates independently and have a unique id, monitoring soil and weather conditions in real-time.

[0056] Slave Node is an ESP32 microcontroller, which connects to several sensors. These include a capacitive soil moisture sensor to assess soil wetness, a DHT22 temperature and humidity sensor to measure the surrounding climate. This data helps the system determine when the crops require watering.

[0057] The Slave Node operates on a solar panel and rechargeable li-ion battery along with BMS, making it suitable for outdoor use without the need for a power line. To conserve energy, the ESP32 wakes up at fixed intervals to take sensor readings, send data to the Master Node, and then returns to deep sleep mode to save power.

[0058] Each node uses the ESP-NOW protocol for communication with Master, which allows the Slave Node to send data directly to the Master Node without needing Wi-Fi or an internet connection.

[0059] Cloud implementation and user interface:

[0060] The system is primarily designed to function offline; however, it also offers optional support for cloud-based data logging and user interaction.

[0061] When Wi-Fi or cellular internet is available, the Master Node uploads sensor data and irrigation logs to the cloud.

[0062] User Interface enables farmers to view real-time data, adjust thresholds, and manually override automation if necessary.

[0063] FIG. 2 illustrates exemplary flow chart of smart irrigation system, in accordance with an embodiment of the present disclosure. The flowchart in FIG. 2 outlines the operational workflow of the intelligent irrigation system, detailing the sequence of initialization, data acquisition, decision-making, and actuation. The process begins with the initialization of the ESP32 master device, which includes the activation of the Real-Time Clock (RTC) module. This clock is essential for synchronizing periodic operations across the entire network. The master device optionally connects to the cloud via web or cellular networks to upload time-stamped data or receive system updates. However, cloud connectivity is not mandatory for core decision-making, as the system is designed to operate autonomously on the edge using local computation.

[0064] At predefined time intervals, as determined by the RTC, the master triggers the data collection phase. This initiates the wake-up process of each ESP32-based slave device assigned to individual crop zones. Each slave device initializes and activates its embedded environmental sensors to collect soil and atmospheric data. These sensors may include, but are not limited to, soil moisture sensors, temperature sensors, humidity sensors, and other relevant components depending on the crop requirements. Once the data is collected, each slave node establishes a connection with the ESP32 master device using the ESP-NOW communication protocol, which enables direct and energy-efficient peer-to-peer data transfer without needing a conventional Wi-Fi router. The collected sensor data is then transmitted to the master device.

[0065] Upon receiving the sensor data from all connected slave nodes, the master device proceeds with localized data analysis and decision-making. The logic for irrigation control is based on evaluating whether the current water levels in each crop zone falls below predefined thresholds. If the water level or soil moisture content in a particular zone is determined to be low, the system initiates the irrigation sequence for that zone. Specifically, the valve associated with the crop zone is opened, and the water pump is activated to begin the water flow. The system continues to monitor the water application process to ensure that the required moisture level is achieved.

[0066] As part of the control logic, once the adequate moisture level is restored in the respective crop zone, the system automatically closes the valve and turns off the water pump. This feedback-controlled operation ensures that each crop receives only the necessary amount of water, thereby avoiding over-irrigation or wastage. The decision-making and actuation processes repeat for each monitored crop zone based on its sensor readings. The loop terminates once all zones are serviced as needed, and the system returns to standby mode, awaiting the next scheduled wake-up cycle.

[0067] The entire workflow as depicted in FIG. 2 demonstrates a fully autonomous, sensor-driven irrigation system with decentralized data acquisition and centralized decision-making. The use of RTC for time management, low-power ESP32 microcontrollers for edge computing, and ESP-NOW for wireless communication ensures high energy efficiency and operational reliability. Moreover, the flowchart highlights the modularity of the system, where new zones or sensors can be integrated seamlessly without reconfiguring the overall architecture.

[0068] This implementation enables precise, demand-driven irrigation control tailored to the unique conditions of individual crop zones. It improves water use efficiency, reduces manual labor, and enhances the sustainability of agricultural practices. The inclusion of conditional cloud connectivity also allows remote monitoring, historical data analysis, and further integration with larger smart farming platforms. Overall, FIG. 2 encapsulates the system’s core operational logic in a robust and technically coherent manner, supporting the invention’s objectives of autonomy, precision, and scalability in smart agriculture.

[0069] Referring again to FIG. 2, the flowchart represents the operations and technical implementation of the irrigation system. It illustrates the following: the initialization process begins with the ESP32 master controller setting up the system and potentially synchronizing with a Real-Time Clock (RTC) module for time-based operations. The data collection architecture employs multiple ESP32 slave units that gather soil moisture readings from various crop fields. For data transmission, the system utilizes the ESP-NOW protocol to send the sensor readings back to the master controller wirelessly. An intelligent decision-making algorithm evaluates the soil moisture levels in each crop field and determines the irrigation requirements. The control system selectively opens valves for specific crops based on their moisture needs and activates the water pump accordingly. Lastly, a continuous monitoring loop checks the reservoir water levels and stops irrigation once the optimal soil moisture is achieved.

[0070] FIGs. 3A-3D illustrates an exemplary experimental set up for the implementation, showing user interface, exemplary crop zone, and mechanical connection between the components, in accordance with an embodiment of the present disclosure.

[0071] An experimental analysis was performed using an engineering experimental setup as shown in FIG. 3D. Following hardware and software components / features were used for implantation:Hardware Requirements:Sr No Name of Component Specification Qty1. ESP32-WROOM-32D Tensilica Xtensa LX6 dual core, Operating Voltage: 3.6V, Operating Current: 500mA, 448 KB ROM, 520 KB SRAM, Wi-Fi + Bluetooth + Bluetooth LE MCU 32. Capacitive Soil Moisture Sensor Operating Voltage: 3.3 ~ 5.5 VDC; Output Voltage: 0 ~ 3.0VDC; Operating Current: 5mA. Interface: PH2.54-3P. 23. DHT22 Digital Temperature and Humidity Sensor Module AM2302 Temperature -40-80 ℃; humidity 0; 99.9%RH, Accuracy: temperature: + 0.5; humidity: + 2%RH (10; 90%RH)Capacitive type 24. DS1307 RTC I2C interface, Hour: Minutes: Seconds AM / PM. Day Month, Date – Year, 1Hz output pin 15. DC Gear motor Voltage (DC):12V, Rated RPM (at 12V):10, Gear Reduction: 810K, Rated Torque (N-cm): 680, Full Load Current (A):2.047 16. Encoder Model: OE-37, Input Supply voltage (V):4.5 ~ 24, Supply Current (mA):14,Output Current (mA):< 0.1 17. Solar Pannel 6v Solar Pannel 38. Li-ion battery Capacity (mAh): 2000; Output Voltage: 3.7V 79. Relay Trigger Voltage: 5 VDC, Trigger Current: 20 mA, Maximum Switching Voltage: 250VAC@10A; 30VDC @10A 110. Water Valve Three Way PVC Ball Valve 1 / 2 211. Water Pump Working voltage: DC 12V, current work: 0.5-0.7A, Maximum suction: 1-2L / Min 112. Pipe ½ inch -Software Requirements:Sr No Name of Software Use Case1. Proteus Design Suite Circuit and PCB Designing2. Arduino IDE Code writing3. Visual Studio Code User interface using HTML, CSS, Bootstrap4. Fusion 360 3D designing

[0072] FIG. 3A illustrates an exemplary user interface when the slave devices send data to the master device to decide whether to initiate the irrigation or not, and if to start how much would be requirements of each zone. FIG. 3B illustrates Crop 1 (Mint(pudina)) in field 1 (zone 1). FIG. 3C illustrates Crop 2 (Coriander) in field 2 (zone 2).

[0073] FIG. 4 illustrates the hardware architecture (400) of a computing system suitable for implementing the ESP32 master device (510) within the autonomous irrigation system (500). The architecture is centered around a system bus (420) that interconnects all key components, ensuring synchronized data flow and control execution. At the heart of the system lies the processor (470), responsible for executing machine learning models, decision algorithms, and communication protocols such as ESP-NOW and Wi-Fi. It operates in coordination with the main memory (430), which provides dynamic storage for real-time sensor data, irrigation states, and control outputs. Firmware, pre-trained inference models, and irrigation routines are stored in the read-only memory (440), ensuring persistent functionality. A mass storage device (450) is included for storing historical data logs, large datasets, and updateable configuration files, typically via embedded flash or SD card. Communication ports (460) interface the processor (470) with ESP32-based slave nodes (502-1 to 502-N) and external cloud services, supporting both local and remote data exchange. This modular and embedded architecture enables reliable, low-power, and scalable operation, forming the computational backbone for zone-specific irrigation decisions and actuator control.

[0074] FIG. 5 illustrates a smart irrigation system (500) configured for adaptive and autonomous multi-zone water management in agricultural environments. The system is architected around a central ESP32 master device (510), which communicates bidirectionally with a plurality of distributed ESP32 slave devices (502-1, 502-2, 502-3, ..., 502-N), each assigned to monitor and control an associated crop zone (504-1, 504-2, 504-3, ..., 504-N). These slave devices are deployed within their respective zones and are responsible for collecting critical environmental and soil condition data using a set of integrated sensors (506).

[0075] Each slave device (502) houses a plurality of sensors (506) that enable high-resolution field monitoring. These include: a capacitive soil moisture sensor (506-1) to detect volumetric water content in the soil, a temperature and humidity sensor (506-2) for atmospheric condition sensing, a soil electrical conductivity (EC) sensor (506-3) to assess salinity levels, an image sensor (506-4) for capturing visual canopy features, and a soil pH sensor (506-5) to monitor acidity or alkalinity relevant to crop health.

[0076] Sensor readings from each slave are managed by a corresponding first microcontroller unit (508-1, 508-2, ..., 508-N). These microcontrollers are typically implemented using ESP32 chips, selected for their low power consumption and wireless communication capabilities. To optimize energy usage, each microcontroller operates primarily in deep sleep mode and is programmed to wake up at fixed intervals, activate the sensors (506), and transmit the collected data using the ESP-NOW protocol (512) directly to the master device (510).

[0077] The master device (510) comprises a second microcontroller unit (514), also implemented using an ESP32 microcontroller. Upon receiving sensor data from various zones via ESP-NOW (512), the second microcontroller performs on-device processing using embedded machine learning algorithms, such as a time-series regression model. This model is trained on both real-time data and historical patterns to determine irrigation requirements tailored for each crop zone. Image data from the image sensor (506-4) is concurrently processed using a machine vision model to assess parameters like leaf area index, canopy wilting, or crop stress. The image analysis is then correlated with the sensor values to generate a comprehensive irrigation control signal.

[0078] Each crop zone (504) is further equipped with a dynamic valve assembly (516-1, 516-2, ..., 516-N), which physically controls the irrigation flow. This assembly consists of a gear-driven three-way irrigation valve (518-1, 518-2, ..., 518-N) for precise channeling of water and a dedicated controller (520-1, 520-2, ..., 520-N). The controller may use a Proportional-Integral-Derivative (PID) control loop or an AI-based feedback mechanism to dynamically modulate the valve's position based on the irrigation control signal sent from the master. This dynamic response ensures accurate, zone-specific water delivery, avoiding both under- and over-irrigation.

[0079] In addition to the control logic, the system may also comprise a water intake module in each crop zone. This module includes a plurality of inlet valves, each connected to distinct water sources such as borewells, rainwater reservoirs, or treated wastewater outlets. A source selection logic operates in real time to switch between these inputs based on measured EC and pH data or reservoir availability. The logic prioritizes sources with optimal water quality characteristics (such as low EC and balanced pH) suitable for the specific crop in that zone.

[0080] The water from the selected source is then delivered to the irrigation zone via a pump, which may be electrically actuated and controlled by the master device (510). Once the necessary moisture levels are achieved in a given zone, as detected via the sensor feedback, the system automatically closes the valve and deactivates the pump, completing the irrigation cycle efficiently.

[0081] For image analysis, the image sensor (506-4) captures photos at predefined intervals, and the master device (510) computes vegetation indices such as the Normalized Difference Vegetation Index (NDVI) or quantifies leaf wilting using predefined image scoring metrics. These indices, correlated with environmental data, serve as crucial decision parameters for determining crop stress and irrigation urgency.

[0082] The entire system operates in a decentralized but coordinated fashion, with the slave devices (502) responsible for field-level sensing and the master device (510) orchestrating centralized logic, prediction, and control. The ESP-NOW protocol (512) is crucial to maintaining lightweight, peer-to-peer, low-power communication without dependency on external Wi-Fi routers or internet infrastructure. This makes the system scalable and reliable, particularly in off-grid or rural installations.

[0083] Working example: Consider an agricultural field divided into four crop zones: 504-1 to 504-4, each monitored by a dedicated ESP32 slave device (502-1 to 502-4). At a scheduled interval (e.g., every 30 minutes), the real-time clock (RTC) in the master device (510) triggers a synchronized data collection cycle. Each slave device wakes from deep sleep mode and activates its attached sensors (506-1 to 506-5). The sensors gather the following readings from crop zone 504-2:

[0084] Soil moisture sensor (506-1): 14% volumetric water content (VWC); Temperature and humidity sensor (506-2): 34°C ambient temperature, 42% relative humidity; Soil EC sensor (506-3): 2.3 dS / m; Soil pH sensor (506-5): pH 6.7; Image sensor (506-4): Captures leaf images showing mild wilting and reduced canopy coverage.

[0085] These sensor values are packaged and transmitted via ESP-NOW protocol (512) to the ESP32 master device (510). The master’s microcontroller unit (514) receives the data and feeds it into an on-device machine learning model—a time-series regression model trained on historical moisture data, weather patterns, and crop type. Based on the input, the model determines that the soil moisture threshold for the crop in zone 504-2 should be maintained between 18% and 28% VWC. The current measured value of 14% is below this range, indicating a need for irrigation.

[0086] Simultaneously, the master device analyzes the image data using a lightweight vision model. The model computes a leaf wilting score of 0.62 (on a scale of 0 to 1) and a normalized difference vegetation index (NDVI) of 0.41, both of which indicate mild crop stress. The system correlates this canopy stress with the low moisture reading and concludes that irrigation must be activated immediately for crop zone 504-2.

[0087] In response, the master device sends an irrigation control signal to the corresponding dynamic valve assembly (516-2). The controller (520-2), equipped with a PID control algorithm, actuates the gear-driven three-way valve (518-2) to open the flow path for water. Simultaneously, the master switches on the water pump, drawing water from a reservoir selected by the source selection logic. This logic prioritizes a source with EC = 1.1 dS / m and pH = 6.8, ideal for the specific crop's tolerance.

[0088] The system continues watering while the slave device (502-2) periodically samples the soil moisture in real time. Once the moisture level reaches 21% VWC, which falls within the acceptable range, the controller receives feedback and begins closing the valve using the PID algorithm to avoid overshooting. The master device then turns off the water pump, and a new data snapshot is recorded for monitoring.

[0089] In another crop zone, say 504-1, the sensor data indicates a moisture level of 23%, NDVI of 0.62, and no signs of wilting. The model predicts no immediate need for water replenishment. Consequently, the irrigation system remains off for that zone, conserving water.

[0090] FIG. 6 illustrates an exemplary method for adaptive and autonomous multi-zone water management in agricultural fields, in accordance with embodiments of the present disclosure. The method for managing irrigation autonomously across multiple crop zones using a network of intelligent sensing and control devices is provided.

[0091] The method initiates with the retrieval (602) of sensor data from a plurality of sensors embedded in ESP32-based slave devices, each assigned to a specific crop zone. These sensors include a capacitive soil moisture sensor, a temperature and humidity sensor, a soil electrical conductivity (EC) sensor, an image sensor, and a soil pH sensor. Each of these sensors continuously or periodically monitors critical environmental and agronomic parameters within its crop zone. For instance, the soil moisture sensor assesses water content in the root zone, while the EC and pH sensors help infer salinity and acidity levels that influence nutrient uptake. The image sensor captures real-time photos of the crop canopy to support visual diagnostics such as wilting, leaf yellowing, or reduced foliage density.

[0092] Once the sensor data is collected, it is transmitted (604) by a first microcontroller unit embedded in each ESP32 slave device to a centralized ESP32 master device using the ESP-NOW wireless communication protocol. ESP-NOW is chosen for its low-latency, low-power, and infrastructure-independent communication, which allows the distributed slave nodes to function in remote agricultural areas without requiring a Wi-Fi router or cellular backbone. Each transmission packet encapsulates the latest readings from the five sensors, along with a timestamp and identification for the source crop zone.

[0093] Upon receiving the data, the second microcontroller unit in the master device performs the receiving operation (606). The master controller parses the incoming data streams from multiple slave devices and temporarily buffers them for processing. At this stage, the master device also synchronizes data using its onboard real-time clock (RTC), aligning all sensor values to a uniform temporal reference for meaningful comparison across zones.

[0094] Following this, the processing step (608) begins, where the second microcontroller analyzes the received sensor data using a lightweight, on-device machine learning model. This model may be a time-series regression model or a decision-tree based classifier, trained on historical environmental data, irrigation history, and crop type. The model predicts soil moisture depletion rates and identifies whether a specific crop zone is currently under-irrigated or requires no action. This decision is made in a predictive and context-aware manner, minimizing both overwatering and stress due to insufficient water.

[0095] The system then proceeds to receive and analyze (610) the image data from the image sensor included in the sensor set. The image is processed locally using a machine vision algorithm, which extracts indicators like Leaf Area Index (LAI), Normalized Difference Vegetation Index (NDVI), and wilting scores. These visual indicators serve as proxies for plant stress and health. By performing feature extraction, the system can identify early signs of dehydration or stress, which may not yet be evident in moisture data alone.

[0096] In the correlation and analysis step (612), the master device integrates the processed image data with environmental sensor readings. This fusion of data modalities enables the system to issue a contextually informed irrigation control signal. For example, if soil moisture is slightly low but NDVI and canopy appearance indicate active growth and no visible stress, the system may defer irrigation. Conversely, low moisture and visible wilting would immediately trigger irrigation. The control signal contains a binary or modulated instruction to either activate or inhibit water delivery to the affected crop zone.

[0097] Finally, the system executes the modulation operation (614) via a controller embedded within the dynamic valve assembly. This assembly features a gear-driven three-way irrigation valve, which can redirect water flow from one or more sources to the targeted crop zone. The controller uses a PID control algorithm to finely adjust the valve position, ensuring precise flow control, or may alternatively employ a reinforcement learning-based feedback mechanism that adapts based on previous irrigation effectiveness. This step closes the feedback loop between sensor-based monitoring, machine-driven decision-making, and physical actuation, allowing for continuous, real-time, and localized water management. , C , Claims:CLAIMS:

Claims

WE CLAIM:

1. An irrigation system (500) for adaptive and autonomous multi-zone water management in agricultural fields, the system comprising:a plurality of Espressif 32 (ESP32) slave devices (502-1, 502-3, …., 502-N), each associated with a corresponding crop zone (504-1, 504-2, ……., 504-N), wherein each ESP32 slave device having:a plurality of sensors (506) including a capacitive soil moisture sensor (506-1), a temperature and humidity sensor (506-2), a soil electrical conductivity (EC) sensor (506-3), an image sensor (506-4), and a soil pH sensor (506-5); a first microcontroller unit (508-1, 508-2, ……., 508-N) configured to operate in deep sleep mode and periodically activate to retrieve and transmit sensor data obtained from each of the plurality of sensors (506) to an espressif 32 (ESP32) master device (510) via ESP-NOW protocol (512); the ESP32 master device (510) having:a second microcontroller unit (514) configured to:receive the transmitted sensor data from the first microcontroller unit;process the received sensor data using on-device machine learning technique trained on real-time and historical sensor data to determine irrigation requirements for each crop zone;receive and analyse image data from the image sensor received in the transmitted sensor data to detect crop canopy features, including stress, wilting, and leaf area index using a machine vision model;analyse and correlate the image data with the environmental sensor readings to generate an irrigation control signal, the irrigation control signal comprising a determination to either activate or inhibit water distribution for the corresponding crop zone; a dynamic valve assembly (516-1, 516-2, ……., 516-N) provided at each crop zone (504-1, 504-2, ……., 504-N) operatively coupled to the master device, the dynamic valve assembly comprising:a gear-driven three-way irrigation valve (518-1, 518-2, ……., 518-N); anda controller (520-1, 520-2, ……., 520-N) configured to modulate the valve position dynamically using a Proportional-Integral-Derivative (PID) controller or an AI-based feedback mechanism for zone-specific water distribution based on the generated irrigation control signal.

2. The system as claimed in claim 1, wherein the system comprises a water intake module provided at each crop zone (504-1, 504-2, ……., 504-N), the water intake module having: a plurality of inlet valves, each connected to a different water source; and a selection logic configured to switch between said water sources based on real-time EC and pH sensor data or reservoir availability, and the different water source selection logic is configured to prioritize water with optimal electrical conductivity and pH ranges for a specific crop zone.

3. The system as claimed in claim 2, wherein the system comprises a pump configured to deliver water from the selected source to the selected irrigation zone.

4. The system as claimed in claim 1, wherein the machine learning algorithm is a time-series regression model trained to predict soil moisture depletion based on environmental inputs and historical irrigation patterns.

5. The system as claimed in claim 1, wherein the image sensor captures images at predefined intervals, and the master device computes vegetation indices such as Normalized Difference Vegetation Index (NDVI) or leaf wilting scores.

6. The system as claimed in claim 1, wherein the plurality of ESP32 slave devices operate in deep sleep mode, and wherein each ESP32 slave device is configured to wake at fixed intervals, transmit data, and return to deep sleep mode.

7. The system as claimed in claim 1, wherein the first microcontroller unit and the second microcontroller unit is ESP32 microcontroller.

8. An espress if 32 (ESP32) master device (510) for adaptive and autonomous multi-zone water management in agricultural fields, the master device comprising:receive sensor data;analyse and correlate the image data with the environmental sensor readings to generate an irrigation control signal, the irrigation control signal comprising a determination to either activate or inhibit water distribution for the corresponding crop zone.

9. The master device as claimed in claim 8, wherein the sensor data is received from a plurality of sensors (506) including a capacitive soil moisture sensor (506-1), a temperature and humidity sensor (506-2), a soil electrical conductivity (EC) sensor (506-3), an image sensor (506-4), and a soil pH sensor (506-5), the plurality of sensors are embedded in each ESP32 slave device selected from the plurality of Espressif 32 (ESP32) slave devices (502-1, 502-3, …., 502-N), each associated with a corresponding crop zone (504-1, 504-2, ……., 504-N); andtransmit the irrigation control signal to a dynamic valve assembly (516-1, 516-2, ……., 516-N) provided at each crop zone (504-1, 504-2, ……., 504-N), the dynamic valve assembly comprising:

10. A method for adaptive and autonomous multi-zone water management in agricultural fields, the method comprising:retrieving (602) sensor data obtained from a plurality of sensors including a capacitive soil moisture sensor, a temperature and humidity sensor, a soil electrical conductivity (EC) sensor, an image sensor, and a soil pH sensor, the plurality of sensors are embedded in each ESP32 slave device selected from a plurality of Espressif 32 (ESP32) slave devices;transmitting (604), by a first microcontroller unit embedded in each ESP32 slave device, the retrieved sensor data to an Espressif 32 (ESP32) master device via ESP-NOW protocol;receiving (606), by a second microcontroller unit of the ESP32 master device, the transmitted sensor data from the first microcontroller unit;processing (608), by the second microcontroller unit, the received sensor data using on-device machine learning technique trained on real-time and historical sensor data to determine irrigation requirements for each crop zone;receiving and analysing (610), by the second microcontroller unit, image data from the image sensor received in the transmitted sensor data to detect crop canopy features, including stress, wilting, and leaf area index using a machine vision model;analysing and correlating (612), by the second microcontroller unit, the image data with the environmental sensor readings to generate an irrigation control signal, the irrigation control signal comprising a determination to either activate or inhibit water distribution for the corresponding crop zone;modulating (614), by a controller of a dynamic valve assembly, a position of a gear-driven three-way irrigation valve dynamically using a Proportional-Integral-Derivative (PID) controller or an AI-based feedback mechanism for zone-specific water distribution based on the generated irrigation control signal.