New Strawberry Picking Robot System Enhances Agricultural Automation
In a significant advancement for agricultural robotics, a team of engineers from Jiangsu University of Science and Technology and Beijing University of Posts and Telecommunications has developed a highly efficient control system for autonomous strawberry harvesting. The innovation, detailed in a peer-reviewed study published in Packaging and Food Machinery, introduces a robust integration of machine vision, real-time feedback control, and remote monitoring to address longstanding challenges in robotic fruit harvesting—namely, accurate fruit recognition, terrain adaptability, and operational stability in unstructured field environments.
The research, led by Dr. Qunbiao Wu, Associate Professor at Jiangsu University of Science and Technology, presents a comprehensive control architecture tailored specifically for ridge-cultivated strawberries—a common farming method in China and increasingly adopted globally. Unlike traditional harvesting robots that struggle with uneven terrain and variable lighting, this new system demonstrates improved navigation accuracy, precise fruit detection, and seamless integration with data-driven farm management tools.
As global agriculture faces mounting pressure from labor shortages, rising production costs, and climate variability, automation in specialty crops like strawberries has become a focal point for innovation. Strawberries, being delicate and highly perishable, require careful handling and selective harvesting based on ripeness. Manual labor remains dominant in this sector, but it is increasingly difficult to sustain due to seasonal workforce fluctuations and the physically demanding nature of the work. Automated harvesting systems promise consistency, reduced waste, and round-the-clock operation, but their adoption has been hindered by technical limitations in perception, mobility, and decision-making under real-world conditions.
The breakthrough described in the study addresses these challenges through a multi-layered control system that combines hardware precision with intelligent software algorithms. At its core, the robot utilizes a dual K60 microcontroller setup—one dedicated to locomotion and the other to manipulation and sensory processing. This architectural choice allows for parallel processing of navigation and harvesting tasks, minimizing latency and improving system responsiveness.
One of the most critical aspects of any harvesting robot is its ability to move reliably across soft, uneven soil without damaging crops or losing balance. The researchers implemented a cascaded PID (Proportional-Integral-Derivative) control strategy for the robot’s drive system, enabling it to maintain a steady speed and accurate trajectory along narrow ridges. By fusing data from an accelerometer, gyroscope, and wheel encoders, the system continuously adjusts the differential speed of the left and right motors to correct directional drift. This closed-loop control ensures that the robot stays centered on the planting bed, even when encountering minor obstacles or changes in ground firmness.
Unlike conventional PID implementations that prioritize rapid response, the team optimized the control parameters to emphasize stability over speed. This design choice reflects the operational reality of agricultural robotics, where smooth, predictable motion is more valuable than agility. Field tests confirmed that the robot could traverse typical ridge layouts at a consistent pace, with minimal oscillation or deviation—key factors in preventing plant damage and ensuring safe operation near human workers.
Equally important is the robot’s ability to identify ripe strawberries amidst a complex visual background of leaves, stems, and shadows. The system employs an OpenMV camera module equipped with an STM32F765VI ARM Cortex-M7 processor, providing sufficient computational power for on-board image processing. The camera captures RGB images, which are then converted to the HSV (Hue, Saturation, Value) color space—a transformation that enhances color-based segmentation under varying lighting conditions.
The image processing pipeline begins with adaptive thresholding to generate a binary mask highlighting red regions likely to correspond to mature fruit. However, raw masks often contain noise from specular highlights, soil reflections, or immature berries. To improve accuracy, the system applies morphological filtering and connected component analysis to eliminate small artifacts and isolate potential fruit candidates. The final output includes the centroid coordinates and size estimates of detected strawberries, which are transmitted to the main controller for harvesting decisions.
Testing revealed a recognition accuracy exceeding 95% for fully ripe strawberries under controlled lighting. To ensure consistent performance in real-world conditions, the robot is equipped with integrated LED illumination surrounding the camera. This feature mitigates the impact of variable sunlight, ensuring stable image quality throughout the day. The researchers emphasized that lighting control is not merely a convenience but a necessity for reliable machine vision in outdoor environments.
Once a strawberry is identified, the robot initiates a sequence of mechanical actions to harvest the fruit. The end effector consists of four articulated arms, each driven by multiple servo motors. The control system calculates the optimal approach path based on the fruit’s position and orientation, then commands the arms to grasp the berry gently while avoiding contact with surrounding foliage. A pneumatic actuation system assists in detaching the fruit from the stem, minimizing bruising and preserving post-harvest quality.
The transportation subsystem moves harvested strawberries to a collection bin using a conveyor mechanism. This component was also subject to PID tuning to ensure smooth acceleration and deceleration, preventing fruit damage during transfer. Data from the conveyor’s motor encoder is fed back into the control loop, allowing the system to adjust speed dynamically based on load and mechanical resistance.
Beyond the physical harvesting process, the research introduces a supervisory monitoring interface developed in MATLAB, enabling real-time oversight of both the robot’s status and environmental conditions. Using Bluetooth serial communication, the robot transmits data including temperature, humidity, and harvest statistics to a remote computer. The monitoring software visualizes this information through interactive charts and provides actionable recommendations based on optimal growing conditions for different stages of strawberry development.
For example, if the ambient humidity falls below the ideal range during fruit maturation, the system alerts the operator and suggests corrective measures such as adjusting irrigation or ventilation. Similarly, the number of ripe and semi-ripe berries harvested per unit time is tracked, offering insights into crop yield and harvest timing. This level of integration positions the robot not just as a labor-saving device but as a node in a broader precision agriculture network.
The development process followed a rigorous engineering methodology, beginning with system-level design, followed by modular hardware and software implementation, and culminating in extensive field testing. The electrical architecture includes dual voltage regulation circuits: LM2596 modules supply high-current power to servos and motors, while MIC5209 regulators provide stable 3.3V output for microcontrollers and sensors. This separation ensures clean power delivery and reduces electromagnetic interference, which is critical for reliable sensor operation.
All circuit boards were manually assembled and subjected to functional testing before integration into the robotic platform. The use of industrial-grade components, such as the L298 dual H-bridge motor driver, enhances durability and fault tolerance in harsh agricultural environments. The researchers noted that mechanical robustness and electrical reliability were prioritized throughout the design phase to support long-term deployment.
The study’s findings have implications beyond strawberry farming. The control strategies and sensing techniques could be adapted for other soft fruits such as raspberries, blueberries, or tomatoes, which face similar harvesting challenges. Moreover, the modular design allows for future upgrades, such as integrating AI-based deep learning models for improved classification or adding GPS and LiDAR for autonomous navigation over larger fields.
From a commercial standpoint, the system represents a step toward economically viable agricultural robotics. While initial costs remain high, the researchers argue that the total cost of ownership will decrease as component prices fall and deployment scales up. Furthermore, the ability to operate during off-peak hours and reduce post-harvest losses can improve farm profitability even with moderate adoption rates.
The work also contributes to the growing body of knowledge in agri-robotics, particularly in the area of edge computing for real-time decision-making. By performing image processing and control computations locally, the robot reduces reliance on cloud connectivity, which may be unreliable in rural areas. This decentralized approach enhances system autonomy and responsiveness, making it suitable for remote or disconnected farms.
Ethical and social considerations were also acknowledged by the research team. While automation may displace some manual labor, it also creates new opportunities in robot maintenance, data analysis, and system supervision. The researchers advocate for inclusive technology development that considers the needs of smallholder farmers and promotes equitable access to advanced tools.
In conclusion, the strawberry harvesting control system developed by Wu, Xu, Zhang, Cai, and Xu represents a significant milestone in agricultural automation. It successfully integrates perception, mobility, manipulation, and data analytics into a cohesive platform capable of operating in real-world farming conditions. The system’s high recognition accuracy, stable locomotion, and intelligent monitoring capabilities demonstrate the maturity of robotic solutions for specialty crops.
Future work will focus on increasing harvesting speed, expanding the robot’s operational range, and validating performance across different strawberry varieties and growing systems. The team is also exploring collaboration with agricultural equipment manufacturers to transition the prototype into a commercially available product.
As the global food system continues to evolve, innovations like this underscore the role of engineering in building resilient, sustainable, and productive agricultural practices. The integration of robotics into farming is no longer a futuristic concept but a tangible reality, and this study provides a clear roadmap for how such systems can be designed, implemented, and validated for real-world impact.
Qunbiao Wu, Kanwen Xu, Hongyuan Zhang, Lihua Cai, Chao Xu. New Control System Enables Autonomous Strawberry Harvesting. Packaging and Food Machinery. DOI: 10.3969/j.issn.1005-1295.2021.02.011