The Plant Whisperer: AI‑Powered Conversations with Your Green Companions

Description

The project involves building a Smart Indoor Farming (SIF) framework where IoT sensors monitor real-time variables such as humidity, temperature, CO₂, soil moisture, and visual plant features. These “plant whispers” are processed by edge computing devices using deep learning models for tasks like disease detection, stress analysis, and growth prediction. We face a number of challenges in implementing the proposed system, such as developing real-time, low-latency communication between sensor networks and actuators required to overcome wireless range limitations, especially in large-scale greenhouses and smart farms. However, we addressed this by exploring mid-to-long-range wireless solutions like LoRa and Wi-Fi HaLoW. Second, creating lightweight Artificial Intelligence (AI) models capable of performing predictive analytics at the edge required balancing inference speed with computational efficiency. Third, validating the system in two distinct climates (Kuwait’s desert and Hungary’s temperate zones) posed additional design and adaptation complexities. The multidisciplinary nature of the project, spanning hardware integration, wireless networks, and AI modelling, also required careful coordination between academic and field experts.

Outcomes

We have made demonstrable progress across multiple implementation phases. A testbed indoor farm has been developed with sensors, actuators, and high-resolution imaging systems, all connected via edge-enabled controllers at Kuwait College of Science and Technology (KCST). In this test bed, we monitor real-time plant health with deep-learning models which further classify leaf stress, petal color variation, and disease presence. Our wireless communication modules have been successfully tested for both short and mid-range data transmission. In collaboration with the Hungarian University of Agriculture and Life Sciences has led to robust cross-validation of sensor data under varied environmental conditions. The results from the current implementation show enhanced precision in irrigation and energy use, contributing to a 30–40% reduction in resource waste. Furthermore, AI inference latency has been minimized through edge processing, enabling autonomous, real-time decision-making without reliance on cloud connectivity. Also, Technology transfer discussions are underway with regional agritech incubators, highlighting the project’s innovation and commercial potential. A Birdseye overview of the proposed project is shown in Figure 1.

Figure 1. The proposed design integrates sensors with communication technology.

The information from the plants present in the above indoor farm is communicated via two LoRa Gateways, enabling long-range, low-power communication. In the project, we connect the Gateway A, as shown in Figure 2, to a machine learning-enabled edge server for real-time processing and predictive analytics. The data from both gateways is sent over the internet to an application server, which manages user access. Similarly, users can interact with the system through graphical interfaces on smartphones and web browsers, allowing remote monitoring, control, and visualization of environmental conditions and system status.

Figure 2. Communication with plants via LoRa gateways.

We also proposed an event-driven architecture, where the systems rely on quick, automated responses to changing environmental conditions. Instead of waiting for scheduled checks or manual input, the proposed event-driven architecture allows devices to react instantly when specific conditions are met. For example, if the temperature rises above a certain level or soil moisture drops too low, the system can trigger an irrigation pump or activate a cooling fan without human intervention. This approach improves efficiency, saves resources, and ensures plants get the necessary care at the right time. The following figure shows how this event-driven logic works in our smart farming setup using LoRa communication, microcontrollers, and a cloud-connected IoT platform. We proposed this event-driven architecture, and it is shown in Figure 3.

Figure 3. Event-driven smart farming system using LoRa to automate actions based on sensor thresholds.

In Figure 4, initial results are obtained by connecting humidity, soil moisture, temperature, and illumination level in the indoor smart farm prototype at KCST.

Figure 4. Data obtained from various sensors plugged into the smart farm at KCST.

Impact on the Community

By making plants' needs audible and actionable, “The Plants Whisperer” project creates a new paradigm in regional agriculture. In Kuwait, it empowers climate-resilient indoor farming, dramatically reducing dependency on food imports while conserving water and energy. It offers local, consistent food production amidst increasing climate extremes in the region. The social impact includes empowering farmers with real-time mobile interfaces, enabling faster, data-informed decisions, and promoting food security at the grassroots level. In addition, economic outcomes include job creation in agri-tech, reduced reliance on volatile imports, and price stabilization for communities. The “Plant Whisperer” initiative delivers significant social, economic, and environmental impact. For Kuwait, it offers a viable alternative to import-dependent agriculture by enabling year-round, climate-resilient indoor farming. In the Hungary, it supports consistent local food production despite rising climate volatility. Both regions benefit from reduced water usage, optimized resource allocation, and early disease prevention, leading to higher yields with lower costs. The project empowers local farmers with mobile-based dashboards for intuitive farm control, bridging the digital divide. It also opens new employment pathways in AI-integrated agriculture, fostering regional innovation. Ultimately, the initiative strengthens food sovereignty, boosts local economies, and sets a replicable model for sustainable smart farming globally.