Editorial: Facility Agriculture Robots and Autonomous Unmanned Management for Crops I

1. Introduction

Due to both continued global population growth and an accelerating aging trend in the rural labor force, agricultural production is facing unprecedented challenges. According to forecasts by the Food and Agriculture Organization of the United Nations, global food production needs to increase by approximately 70% by 2050 to meet the expected demand [1]. However, traditional agricultural production methods are not only constrained by rigid limits on land and water resources but are also becoming increasingly vulnerable due to labor shortages, rising production costs, and frequent extreme weather events caused by climate change [2]. Against this backdrop, the transformation of agriculture from experience-dependent to data-driven, as well as from human-powered to intelligent and autonomous, is no longer an academic hypothesis but a practical and urgent necessity for ensuring global food security [3].

Facility agriculture, as an intensive, controllable, and efficient agricultural production paradigm, offers an important pathway for overcoming natural constraints and achieving year-round stable production. However, the high temperature and humidity, limited space, and repetitive operational characteristics of facility environments also impose higher demands on manual management [4]. Concurrently, the rapid development of autonomous unmanned management technologies is providing systematic solutions to this dilemma. The deep integration of robotics, artificial intelligence, the Internet of Things, edge computing, and advanced sensing systems is reshaping how crops are cultivated effectively, thereby driving agriculture via an evolutionary leap from mechanization to intelligence and from automation to autonomy [5,6].

At the technical level, deep learning-based object detection algorithms have achieved significant breakthroughs in agricultural visual recognition in recent years, enabling robots to accurately identify fruits, flowers, weeds, pests, and diseases in complex environments [7,8,9]. At the same time, the maturation of autonomous navigation and path planning technologies allows unmanned ground vehicles and aerial drones to operate efficiently in both row-crop and unstructured scenarios [10]. The fusion of multi-source remote sensing data with meteorological and agronomic variables provides unprecedented perceptual capabilities for crop nutritional diagnosis and precise water and fertilizer management [11]. Furthermore, research in areas such as robotic arm control [12,13], precision seeding [14], greenhouse environment regulation [15], and economic feasibility assessments is also advancing rapidly. The integration and synergy of these technological elements are progressively constructing a closed loop for agricultural unmanned management that integrates perception, decision-making, and execution [16].

To systematically showcase the latest research achievements in this field and to promote academic exchange and technological implementation, we establish the Special Issue “Facility Agriculture Robots and Autonomous Unmanned Management for Crops”. This Special Issue aims to gather research on key technologies spanning the entire chain, from environmental perception and intelligent decision-making to precision operations, with a focus on the effective interface between algorithmic innovation, system integration, and engineering application.

This Special Issue comprises 12 high-quality papers that have undergone rigorous peer review, covering a wide range of application scenarios from field crops (such as rice, maize, and rapeseed) to horticultural crops (such as kiwifruit, strawberries, cherries, and pineapple). The research themes encompass autonomous navigation and path planning, deep learning-based fruit and defect detection, UAV remote sensing for crop nutrient monitoring, robotic arm vision guidance and path optimization, performance evaluation of seeding and harvesting equipment, and intelligent greenhouse environment regulation, among other directions. These studies not only reflect the current technological frontier in the field of facility agriculture robotics and unmanned management but also provide valuable theoretical support and practical references for the development, testing, and large-scale application of future intelligent agricultural equipment.

Through this Special Issue, we hope to present readers with a developmental picture of agricultural automation, from algorithms to equipment and from individual intelligence to systemic collaboration, and to inspire more interdisciplinary scholars to engage in this promising and challenging research field.

2. Overview of Published Articles

The 12 articles included in this Special Issue, centered around the core themes of facility agriculture robotics and autonomous management, can be broadly summarized into the following research hotspots:

2.1. Deep Learning-Based Object Detection and Recognition in Complex Environments

Accurate object recognition is a prerequisite for robots to perform automated operations such as picking, pollination, and weeding. Several papers focus on improving algorithm robustness and lightweight performance under challenging conditions involving occlusion, illumination variations, and dense scenes.

To address the issue of poor model adaptability in complex kiwifruit orchard environments, Zhou et al. proposed the Kiwi-YOLO algorithm. By improving the backbone network of YOLOv8, this algorithm effectively tackles challenges such as variable lighting, severe occlusion, and dense fruit clusters, achieving high-precision kiwifruit detection. Teng et al. developed a lightweight detection framework, DS-YOLO, specifically designed to solve difficulties in strawberry detection due to leaf occlusion, fruit overlap, and illumination changes. This framework maintains accuracy while reducing computational complexity, making it suitable for deployment on mobile terminals. In the context of cherry applications, Ren et al. proposed the SMD-YOLO model, focusing on cherry blossom detection and providing a visual basis for optimizing parameters in intelligent spraying and pollination operations. Furthermore, Liu et al. utilized an improved YOLOv8-DCPF method to achieve rapid and accurate detection of surface defects and malformations in fresh sweet cherries, which is crucial for intelligent sorting systems to enhance the economic value of the produce. Wang et al., who was targeting the issue of high occlusion rates in pineapple harvesting, proposed the Yolov10n-SSE model, which integrates split attention and channel features, along with a hybrid algorithm to optimize robotic arm path planning, demonstrating a complete technical chain from vision to action.

2.2. Autonomous Navigation and Field Path Planning

The autonomous operational capability of robots is highly dependent on reliable navigation technology. Yu et al. filled a gap in research on autonomous navigation for weeding robots in non-row-crop turf environments. They developed a navigation system based on a wheeled unmanned ground vehicle, providing an efficient solution for precision weeding in unstructured scenes like turf. The research by Liu et al. focused on extracting the centerline of corn seedling bands in conservation tillage by proposing an enhanced multi-threshold Otsu algorithm. This algorithm effectively handles image segmentation problems under complex field conditions, such as those with straw row mulching, thus providing an accurate navigation benchmark for seeding and spraying operations.

2.3. Non-Destructive Monitoring of Crop Physiological Parameters and Precision Management

Using UAVs and other platforms for high-throughput crop monitoring is key to achieving precise water and fertilizer management. Qiu et al. overcame the limitations of traditional estimation models that rely solely on crop phenotyping. By integrating UAV imagery with meteorological and fertilization variables, they significantly improved the estimation accuracy and stability of the rice nitrogen nutrition index. Sun et al. focused on monitoring chlorophyll content in rapeseed. By fusing features from UAV-mounted multispectral and RGB imagery, they established a cross-variety general prediction model, demonstrating the significant potential of multi-source data fusion in precision agricultural management.

2.4. Agricultural Machinery Innovation and Operational Performance Evaluation

Beyond intelligent software algorithms, the mechanical innovation and operational performance of agricultural equipment itself are equally important. To address the issues of poor quality and low efficiency in direct rice seeding, Shi et al. developed a double-press precision depth-control seeding technology and optimized its parameters using the discrete element method, providing new equipment for simplified rice cultivation. Hwang and Kim conducted a systematic field performance and economic feasibility assessment of mechanized onion production in South Korea by analyzing the mechanical performance of transplanting, stem cutting, harvesting, and collection processes, offering practical data support to address rural labor shortages. Additionally, Han et al. focused on greenhouse environment control by designing and evaluating a novel thermoelectric-based dehumidifier. This device is particularly suitable for greenhouses in cold regions, as it effectively reduced the impact of high humidity on crops while enabling energy recovery, reflecting innovative thinking in facility agriculture environmental regulation.

3. Conclusions and Future Perspectives

This Special Issue, titled “Facility Agriculture Robots and Autonomous Unmanned Management for Crops”, has successfully brought together 12 cutting-edge research achievements, comprehensively demonstrating the significant potential of intelligent equipment and algorithms in improving agricultural production efficiency, reducing labor intensity, and achieving precision operations. The articles collected exhibit several notable characteristics and future trends.

Firstly, the development of deep learning is important for agriculture. Specifically, the optimization and application of YOLO-series algorithms remain a dominant paradigm in the field of agricultural visual perception. Multiple papers have effectively improved YOLO for specific crops (e.g., kiwifruit, strawberry, and cherry) and environmental challenges (occlusion, illumination, and deformation) by modifying network structures, attention mechanisms, and lightweight designs, demonstrating that task-driven algorithm customization is a key pathway to enhancing practical application effectiveness.

Secondly, research and development in robotics are evolving from single-function perception towards integrated closed-loop systems encompassing perception, decision-making, and action. The synergistic optimization of visual detection and robotic arm path planning in the pineapple harvesting study, for example, represents a mainstream direction for the future development of agricultural robots. The deep integration of technologies across the entire chain, from environmental information acquisition to precise control of the end-effector, will be central to overcoming bottlenecks in complex operations.

Thirdly, multi-source data fusion and coupled environment–crop-management modeling are emerging as new frontiers in precision agriculture. Whether integrating meteorological and fertilization data to estimate rice nitrogen status or combining multispectral and RGB imagery to monitor rapeseed chlorophyll content, it is evident that relying solely on sensor signals is insufficient to meet high accuracy demands. In the future, integrated decision-making systems that combine remote sensing data, environmental factors, crop models, and agronomic knowledge will play an even more critical role.

Finally, this Special Issue also reveals current research challenges and future research gaps. The generalization ability of algorithms remains a major obstacle limiting the cross-regional and cross-variety application of models. The operational stability of robots under extreme weather conditions and in highly unstructured environments requires substantial further validation. Economic feasibility assessments for the transition from technological development to commercial application (as exemplified by the onion production study) should be integrated earlier into the research design. Furthermore, identifying strategies that reduce the hardware costs of intelligent equipment and simplify user interfaces to better suit agricultural production systems dominated by smallholder farmers is a practical issue that must be addressed for widespread adoption.

In summary, this Special Issue provides readers with a window into the current state of facility agriculture robotics and autonomous management technologies. We believe that these research efforts not only stimulate academic discussion in this field but also provide valuable theoretical foundations and practical references that drive the development and deployment of next-generation intelligent agricultural equipment. We look forward to seeing future breakthrough research on multi-robot collaboration, human–robot collaboration, and the application of new-generation artificial intelligence (such as large models) in agriculture.