Tower-typed hydroponic smart farm using automated harvester

1. Introduction

Agriculture, the foundation of human civilization, plays a critical role in both ecosystems and the economy. It encompasses all activities related to food and livestock production, remaining indispensable for human survival and development. Sustaining the agricultural industry requires a stable workforce. However, modern society is grappling with a severe labor shortage in this sector. However, modern society is facing a severe agricultural labor shortage. In South Korea, declining birth rates and an aging population have accelerated the decrease in agricultural labor, while rural populations continue to decline [1]. As the average age of rural residents rises, the issue of labor shortages intensifies.

To address these challenges, a novel agricultural technology known as smart farming has emerged. Smart farming integrates information and communication technologies (ICT) to automate and optimize agricultural production, management, and distribution processes. This approach not only improves efficiency and productivity but also enables precision farming and real-time decision-making [2]. Smart farms are increasingly recognized for their ability to enhance productivity and reduce labor demands in agriculture. By integrating renewable energy and hydroponic systems, they also address environmental challenges and improve energy efficiency. For example, recent studies have demonstrated the effectiveness of combining solar and wind energy in smart farm systems, achieving significant resource savings while maintaining optimal growth environments for crops [3].

By leveraging advanced technologies such as sensors, artificial intelligence (AI), big data, and the Internet of Things (IoT), smart farming systems enable real-time monitoring of crop growth environments and automatic adjustments of essential resources such as water, nutrients, temperature, and light to maximize crop productivity and quality [4].

Among the various cultivation methods, hydroponics stands out in smart farming for its resource efficiency and environmental sustainability. Compared to soil-based farming, hydroponics offers advantages such as soil erosion prevention, pest and disease control, and optimized water usage [5]. Strawberries are particularly well-suited to hydroponic systems integrated with smart farming technologies. Their distinctive growth stages and clear color transitions make them ideal candidates for automated harvesting systems. Figure 1 illustrates the growth stages of strawberries, from flowering to full ripeness. These stages are essential for implementing the color recognition system within the smart farm. Despite advancements in smart farming, existing systems predominantly focus on growth management, with harvesting remaining largely dependent on manual labor [6]. To address this gap, this study aims to design a hydroponics-based smart farm system specifically for strawberry cultivation, automating both growth management and harvesting processes. The proposed system incorporates a deep learning-based color recognition technology and an automated harvesting algorithm to accurately assess strawberry ripeness and initiate harvesting at the optimal time [7][8]. Through this research, the potential of smart farming systems to reduce labor dependency and improve productivity will be demonstrated, offering viable solutions to challenges faced by South Korean agriculture [9].

Figure 1:

The Growth Process of Strawberries

2. Integrated Components of the Smart Farm

The smart farm system proposed in this study is designed to comprehensively manage the growth, harvesting, and monitoring of strawberries. It comprises three main components: growth management, automated harvesting, and monitoring. Building on these three components, the system's overall structure and harvesting algorithm are designed to integrate their functionalities seamlessly. The smart farm structure optimizes the physical arrangement of components to facilitate efficient operation, while the algorithm ensures synchronized and automated processes for strawberry cultivation and harvesting.

2.1 Growth Management

The growth management component operates independently to optimize the environmental conditions for strawberry growth, focusing on factors like temperature, humidity, and water levels. As shown in Figure 2, the monitoring component tracks the environmental conditions that directly influence the growth of strawberries, such as temperature and humidity, while the automated harvesting system ensures that strawberries are harvested at the optimal time based on their ripeness.

Figure 2:

Smart farm system diagram

The growth management component is designed to optimize and automatically control environmental conditions to ensure stable growth of strawberries. The system provides the following ideal environments for each component in crop development from flowering to fruit formation:

• i) LED Lighting: LED lighting ensures consistent illumination, stimulating photosynthesis and improving growth efficiency. By providing uniform light regardless of weather conditions, it supports the stable growth of strawberries.
• ii) Water Level Sensors and Pumps: Water level sensors and pumps detect water loss in the hydroponic environment and automatically adjust the water level to supply appropriate moisture to the strawberry roots.
• iii) Fans: Fans create airflow around the flowers, mimicking natural pollination processes and aiding in successful fruit formation.
• iv) Temperature sensor: Temperature sensors monitor the water temperature in the cultivation tank in real time, maintaining the optimal growth temperature for strawberries, which is 17–20°C. When temperature changes are detected, the system automatically sends alerts.

This growth management system continuously fine-tunes the conditions necessary for strawberry growth, ensuring efficient and stable development.

2.4 Smart Farm Structure

The smart farm is designed with a three-tier structure to provide optimal environments tailored to each growth stage of strawberries, adjusting environmental conditions for each layer to optimize the growth and harvesting processes (Figure 3). The first tier (Zones E and F) serves as the nursery zone, providing the environment necessary for the early growth stages of strawberries. This zone is dedicated to seedlings prior to flowering, supplying consistent light and moisture to promote healthy root and stem development through stable LED lighting and water pumps paired with level sensors to maintain proper moisture conditions. The second tier (Zones C and D) is the fruit formation zone, supporting the post-flowering stages of strawberry growth. Fans facilitate artificial pollination, promoting successful fruit formation, while optimal temperature and humidity levels are maintained to ensure smooth and robust fruit development. The third tier (Zones A and B) is the harvesting and storage zone, designed for collecting fully ripe strawberries and storing new seedlings. Zone A serves as the storage area for harvested strawberries, while Zone B stores new seedlings to be relocated to the appropriate zones as needed.

Figure 3:

Smart farm modeling

Special zones within this structure enhance functionality. Zones C and E serve as color recognition zones, where the device evaluates the ripeness of strawberries. Once full ripeness is detected, the automated harvesting system is activated to collect the strawberries. Zones D and F act as standby zones for strawberries before they are transferred to the color recognition zones. This integrated tiered structure allows the smart farm to operate efficiently, segmenting the growth and harvesting processes into distinct but harmonized stages that optimize the overall performance.

2.5 Algorithm

The automated harvesting algorithm of the smart farm operates through the integration of the color recognition device and the automated harvester, evaluating the ripeness of strawberries before carrying out harvesting tasks. The algorithm functions at a designated time each day and follows these sequential steps (Figure 4):

Figure 4:

Automated harvesting algorithm diagram

The process begins with the color recognition device moving to Zone C, where it evaluates the ripeness of strawberries in the trays. Once positioned, the device scans the strawberries in Zone C, waiting five seconds to complete the analysis. If a fully ripe (100%) strawberry is detected, the automated harvester is activated to transport the tray from Zone C to Zone A for harvesting, while the tray in Zone D is moved to Zone C. Simultaneously, the color recognition device transitions to Zone F to prepare for the next operation.

If no fully ripe strawberries are detected in Zone C, the color recognition device immediately moves to Zone F, where it evaluates the flowers to determine whether they are in a blooming state (white color). If blooming flowers are detected, the automated harvester is activated to transport the tray from Zone E to Zone D, while the tray in Zone F is moved to Zone E. New seedlings are then retrieved from Zone B and placed in Zone F. Upon completing these tasks, the color recognition device and automated harvester return to their initial positions, ready for the next cycle. If no blooming flowers are detected in Zone F, the color recognition device returns to its initial position, and the automated harvester remains on standby.

This algorithm operates on a 24-hour cycle, ensuring continuous automation of the smart farm's processes and maintaining efficiency and consistency in strawberry harvesting and management.

3. Experimental Setup and Results

3.2 Experimental Results

In this study, multiple experiments were conducted to evaluate the performance of the designed automated harvesting component.

First, the color recognition system integrated into the harvesting system demonstrated an average recognition accuracy of 95.5%, evaluated through 100 repeated attempts per test (N=100) for each growth stage (blooming, 30%, 70%, and 100% maturity). When tested across various color saturations, the recognition accuracy varied by 8% to 28%; however, this variation was within an acceptable error range, confirming that segmented cultivation and automated harvesting based on predefined criteria are feasible within the system. Figure 5 illustrates the recognition performance of the system at different growth stages. While the recognition accuracy for 30% and 70% mature strawberries is moderate, the system achieves a significantly higher accuracy at 100% maturity, as shown by the steep improvement in recognition rates. This trend underscores the system's reliability in identifying fully ripe strawberries, which are the primary target for harvesting.

Figure 5:

Recognition rate of strawberries

Second, to reduce potential shaking and positional errors of the harvester during coordinate-based movement, the tray structure was optimized during the system design process. The accuracy of tray movement was tested with 100 repeated attempts (N=100) across predefined paths, as shown in Figure 6. Activating the automatic harvester recorded an average error rate of 4%, showing that the algorithm enables accurate path movement with minimal system errors.

Figure 6:

Success rate of moving trays

These experimental results support the effectiveness of the proposed tower-type smart farm system in terms of automated harvesting, energy efficiency, and system stability.

4. Conclusion

The tower-typed smart farm system designed in this study has demonstrated its potential as an integrated solution capable of automated harvesting, energy efficiency, and stability. Experimental results confirmed the system's ability to achieve a 95.5% recognition accuracy for strawberry ripeness, and the optimized structure improved crop yield by multiple compared to conventional setups. By integrating renewable energy sources such as solar and wind power, the system addresses the challenges of energy consumption and labor shortages in traditional farming methods. Additionally, the use of a three-tiered structure optimizes space utilization and enhances productivity per unit area.

However, further research is needed to enhance the system's scalability and adaptability to other crops. Incorporating more advanced deep learning models could improve color recognition accuracy, and adding automated crop transportation features could further reduce labor demands. This study highlights the potential of smart farm systems to revolutionize modern agriculture by improving sustainability and reducing dependency on human labor. With continuous advancements, such systems could play a pivotal role in ensuring global food security and addressing climate change challenges.

This research aligns with previous findings by Muangprathub et al. [10], who highlighted the role of IoT and data analysis in enhancing smart farm efficiency. By leveraging IoT-based components and deep learning techniques, this study validates the potential of combining these technologies to automate and optimize agricultural processes. Similarly, Kim et al. [11] emphasized the importance of structural optimization in smart farming, and the tower-typed design presented here addresses spatial constraints while improving productivity.

Furthermore, Ljungberg et al. [12] demonstrated that machine learning applications can significantly enhance agricultural datasets, which parallels the use of the HuskyLens AI module in this study to achieve high accuracy in strawberry ripeness detection. Ryu et al. [13] also showcased the benefits of IoT-enabled automated farming systems, further supporting the practicality of this system in reducing labor dependency and ensuring operational stability.

Lastly, Walter et al. [14] emphasized the role of smart farming in sustainable agriculture, particularly through energy-efficient technologies. The integration of solar and wind energy in this system not only reduces operational costs but also addresses environmental concerns, making it a viable solution for sustainable food production.

First, incorporating additional deep learning techniques for enhanced object recognition could reduce recognition errors caused by interference around the Husky Lens, enabling more accurate and stable monitoring. This would refine the precision of harvest criteria, reduce harvester malfunctions, and contribute to overall productivity improvements. Specifically, precisely timed harvesting based on maturity levels would offer substantial benefits for quality control.

Second, introducing supplementary devices such as water temperature regulators to support the cultivation of a broader range of crops would allow for optimized environmental conditions tailored to specific crops. This would expand the range of cultivable crops and maintain high quality under various conditions. Consequently, the smart farm system's applicability could extend flexibly across diverse agricultural environments, from small farms to large-scale agricultural facilities.

Third, automating the transport of harvested crops by installing a conveyor belt on the third level would further reduce manual labor and enhance the operational efficiency of the smart farm. This improvement would increase processing speed for harvested crops, lower labor costs, and contribute to addressing workforce shortages.

Through these enhancements, the tower-type smart farm system could evolve into a more efficient and reliable automated agricultural solution. This system holds the potential to lead the future of smart agriculture, excelling in energy conservation, productivity, and multi-crop adaptability, and playing a vital role in ensuring a stable food supply and promoting sustainable agricultural practices.

Acknowledgments

This work was supported by the Regional Industry-linked University Open-Lab Promotion Support Project and the KMOU Academic research fund in 2024.

References

• E. S. Mohamed, AA. Belal, S. K. Abd-Elmabod, M. A. El-Shirbeny, A. Gad, and M. B. Zahran, “Smart farming for improving agricultural management,” The Egyptian Journal of Remote Sensing and Space Sciences, vol. 24, no. 3, pp. 971-981, 2021. [https://doi.org/10.1016/j.ejrs.2021.08.007]
• N. Srivastava, R. Kumar, and A. K. Singh, “Technological revolutions in smart farming: Current trends, challenges & future directions,” Computers and Electronics in Agriculture, vol. 198, 106874, 2022.
• H. Kim, I. Kim, J. -W. Seo, and J. Ko, “Smart farm for hydroponic cultivation using integrated renewable energy systems,” IEEE Sensors Journal, 2024. [https://doi.org/10.1109/JSEN.2024.3449344]
• P. Angelov, S. Gupta, and R. Ramezani, “Sensors for smart agriculture: Applications and challenges,” Sensors, vol. 20, no. 15, p. 4231, 2020. [https://doi.org/10.3390/s20154231]
• N. Wahab and R. Hashim, “Optimal plant growth in smart farm hydroponics using IoT and renewable energy,” ASTES Journal, vol. 5, no. 3, pp. 567-574, 2020.
• W. Yang, L. Chen, and X. Zhu, “A review of computer vision technologies for plant phenotyping,” Computers and Electronics in Agriculture, vol. 195, 106773, 2023.
• Y. Park, S. Jeong, and C. Kim, “Smart farming innovations and opportunities,” Agriculture, vol. 13, no. 8, p. 1593, 2023. [https://doi.org/10.3390/agriculture13081593]
• Y. Lin, Q. Wu, and Z. Zhang, “From smart farming towards agriculture 5.0: A review on crop data management,” Agronomy, vol. 12, no. 10, p. 1745, 2022.
• M. Alam, S. Hassan, and M. Jahan, “Application of IoT in smart agriculture,” Proceedings of IEEE International Conference on Advanced Networks and Telecommunications Systems (ANTS), pp. 567-574, 2020. [Online]. Available: https://ieeexplore.ieee.org/document/9108212, .
• J. Muangprathub, N. Boonnam, S. Kajornkasirat, et al., “IoT and agriculture data analysis for smart farm enhancement,” Computers and Electronics in Agriculture, vol. 156, pp. 467–474, 2019. [https://doi.org/10.1016/j.compag.2018.12.011]
• G. J. Kim and J. D. Huhl, “Trends and prospects of smart farm technology,” ETRI Journal, vol. 10, no. 1, pp. 23–35, 2015.
• E. B. Ljungberg, O. Kajdanowicz, and M. Broda, “Machine learning applications on agricultural datasets for smart farm enhancement,” Machines, vol. 6, no. 3, p. 38, 2018. [https://doi.org/10.3390/machines6030038]
• M. Ryu, J. Yun, T. Miao, et al., “Design and implementation of a connected farm for smart farming system,” in Proceedings of the IEEE Sensors Conference, pp. 1–4, 2015. [https://doi.org/10.1109/ICSENS.2015.7370624]
• A. Walter, R. Finger, and R. Huber, “Smart farming is key to developing sustainable agriculture,” PNAS, vol. 114, no. 24, pp. 6148–6150, 2017. [https://doi.org/10.1073/pnas.1707462114]