A Review of Robotic Applications in Livestock and Poultry Farming

 The robots used in livestock and poultry farming can be categorized based on their functions into feeding, cleaning, inspection, and milking robots. The current state and limitations of each type of robot in livestock farming are analyzed as follows:

1. Feeding Robots Traditional feeding in livestock farming primarily relies on manual labor, making it difficult to precisely control feed intake. With the development of modern animal husbandry, automated feeding robots have become widely adopted, improving production efficiency and animal welfare. For example, Nedap's automated feeding systems allow precise management of pigs' feeding behavior. However, the high cost of equipment and the complexity of operation limit their use by small-scale farmers. Additionally, as the demand for precision feeding grows, the complexity of data collection and processing remains a significant challenge.

2. Cleaning Robots Waste removal in livestock farming is a crucial aspect of environmental management. Early mechanical cleaning equipment was inefficient and struggled to distinguish between waste and other materials. With technological advances, self-propelled manure cleaning robots and pigsty cleaning robots have gradually been introduced, but they still face challenges such as high implementation costs and complex maintenance. Smart cleaning robots, using technologies like LiDAR and SLAM algorithms, have improved cleaning efficiency. However, issues with adapting to changing environments and maintenance requirements still need to be addressed.

3. Inspection Robots Smart inspection robots are used in livestock farming for environmental monitoring, animal health management, and data collection. These robots can monitor the temperature, humidity, and gas levels in barns in real-time and use infrared imaging to detect animal body temperature. Inspection robots, such as egg-collecting robots, have achieved automated data collection and counting. Nevertheless, the complexity of data processing and the robots' ability to adapt to different environments remain major hurdles in their development.

4. Milking Robots Automated milking robots have significantly improved milking efficiency and hygiene standards. For example, Sweden's DeLaval has developed a rotary milking robot that efficiently automates the entire milking process. However, high costs and significant maintenance requirements still pose challenges. Researchers are working to improve the flexibility of robotic arms and intelligent control systems, enhancing automation levels, but cost and equipment adaptability remain issues that need to be addressed.

5. Other Types of Robots In addition to the robots mentioned above, modern agriculture has introduced other types of robots, such as sheep shearing robots and egg sorting robots. These technological advancements demonstrate the vast potential for robotics in agriculture. However, the high cost and operational complexity of these technologies continue to hinder widespread adoption.

The application of robots in livestock and poultry farming is driving the industry towards greater efficiency and intelligence. However, cost and technical complexity remain the primary challenges to broader adoption. 

References

[1] HUANG Y, XIAO D, LIU J, et al. Analysis of pig activity level and body temperature variation based on ear tag data[J]. Computers and Electronics in Agriculture, 2024, 219: 108768.
[2] Lü Enli, He Xinyuan, Luo Yizhi, et al. Design of an intelligent feeding IoT system for lactating sows[J]. Journal of South China Agricultural University, 2023, 44(1): 57-64.
[3] Zhu Jun, Ma Shuoshi, Mu Houchun, et al. Design and experimentation of an automatic precise feeding system for breeding pigs[J]. Transactions of the Chinese Society for Agricultural Machinery, 2010, 41(12): 174-177.
[4] Wang Zecheng. Ruibo Le: Witnessing the future of pig farming with technological innovation – An interview with Ruibo Le (Shanghai) Trading Co., Ltd.[J]. Swine Industry Science, 2018, 35(11): 56-59.
[5] Hu Shengjie, Wang Shucai. Application of RFID technology in pig farming[J]. Hubei Agricultural Mechanization, 2007(5): 24-25.

Read More

Intelligent Patrol Robot Path Planning Algorithms

The advent of patrol robots significantly improves the efficiency of patrol tasks, reduces costs, and ensures continuity and stability in operations. Path planning plays a crucial role in the operation of patrol robots, as it must consider factors such as the distance to target points, environmental safety, road conditions, and potential obstacles to determine the optimal patrol route. An excellent path planning algorithm can enhance the efficiency of patrol robots, ensure the safety of the path, and reduce resource and energy waste.

1. Global Planning Algorithms

Common global planning algorithms include Dijkstra's algorithm, A* algorithm, and cost map algorithms. These algorithms aim to optimize the path globally to find the shortest possible route.

(1) Dijkstra's Algorithm

Dijkstra's algorithm calculates the shortest path from the starting point to all other nodes. It dynamically updates the distance information between nodes to choose the next node for expansion until the target node is reached, resulting in the shortest path across the entire map. Dijkstra's algorithm is known for its accuracy and broad applicability, as it can be used for both directed and undirected graphs and can handle edges with weights. However, it has a time complexity of O(V²), where V represents the number of nodes, making it slow for large graphs. Additionally, Dijkstra's algorithm cannot handle negative edge weights, as it assumes all edge weights are non-negative.

(2) A Algorithm*

The A* algorithm is a heuristic global path planning algorithm. It uses a custom heuristic function to estimate the cost between each node and the target, combining this with the actual cost for decision-making. A* ensures both high-quality paths and efficient searches. However, designing an effective heuristic function requires domain knowledge, and the algorithm can encounter memory issues when dealing with large search spaces.

Improved versions of the A* algorithm have been developed by incorporating concepts from potential fields, such as target attraction and obstacle repulsion, into the cost function. Smoothing techniques, like B-spline curves, are also used to create more seamless and realistic paths, which are particularly useful in scenarios where straight-line distances are similar.

(3) Cost Map Algorithm

The cost map algorithm divides the environment into grids, assigning a cost or weight to each grid. The robot can then select the optimal path based on the cost map information. Researchers have extensively studied cost map algorithms, such as the Occupancy Grid Map (OGM) algorithm, which divides the environment into grids and uses binary values to indicate the presence of obstacles. Probabilistic cost map algorithms have also been developed for path planning in uncertain environments. Some studies combine cost maps with the A* algorithm to achieve faster and more optimized path planning.

(4) Voronoi Diagram Algorithm

The Voronoi diagram algorithm divides the environment into regions centered around obstacles, assigning each point in the region to its closest obstacle. Patrol robots can use this algorithm to generate paths that stay within these regions, helping avoid obstacles while maintaining an even distribution of routes. Researchers have enhanced the traditional Voronoi diagram algorithm, especially for multi-robot path planning, by introducing extra guiding points or removing unnecessary feature points to improve path efficiency.

(5) RRT Algorithm

The Rapidly-exploring Random Tree (RRT) algorithm is a path planning algorithm based on sampling and tree structures. It generates a tree by randomly sampling and expanding nodes, gradually approaching the target point. RRT is particularly suitable for high-dimensional, non-convex spaces and has been widely applied in patrol robots' global path planning. Variants such as RRT* and RRT-Connect are also commonly used.

2. Local Planning Algorithms

Local planning algorithms, such as depth-first search, breadth-first search, and dynamic programming, are mainly used to find feasible paths for robots in local environments.

(1) Dynamic Window Approach (DWA)

This approach discretizes the robot's speed and steering space into a limited set of candidate windows. By using local perception data, DWA selects the best combination of speed and steering. It can update the robot's motion state in real-time, adapting to dynamic environments and avoiding collisions. An improved version of DWA considers obstacles both on and near the trajectory to ensure safer navigation.

(2) Model Predictive Control (MPC)

MPC creates a model to predict the robot's movement, then uses optimization to select the best control strategy. It considers the robot's dynamic and target constraints, generating smooth trajectories while avoiding obstacles. Some studies have improved MPC by applying probabilistic models or machine learning techniques, making the algorithm more accurate and stable in complex environments.

(3) Feedback Trajectory Tracking Control

Feedback control is used to adjust the robot's position and trajectory based on predefined paths. By continually modifying control inputs, the robot can adapt to environmental changes and shifting targets. Researchers have developed improved feedback trajectory tracking methods using fuzzy logic control and adaptive control theory, which enhance tracking accuracy and robustness.

(4) Obstacle Avoidance Based on Local Perception

This method uses local perception data (such as LiDAR or camera inputs) to detect nearby obstacles in real-time and apply corresponding avoidance strategies, such as rerouting or stopping. Some studies utilize deep learning to identify and classify obstacles, enhancing the robot's autonomous obstacle avoidance capabilities. Others propose potential field methods to guide the robot's movement and improve obstacle avoidance.

---

References

1. WANG Y, YANG C, ZHOU D. A machine learning-based cost map generation method for mobile robot path planning [J]. IEEE Access, 2022, 10: 55508-55519.
2. KARAMAN S, WALRAND J, FRAZZOLI E. Incremental sampling-based algorithms for optimal motion planning [C]// Proceedings of the IEEE International Conference on Robotics and Automation (ICRA). [S.l.]: IEEE, 2010: 1720-1725.

Read More

Overview of Global Agricultural Robotics Development: Trends, Technologies, and Future Applications

Overview: This article provides a comprehensive review of the concept and characteristics of agricultural robots, examining the development status of agricultural robots in several countries. The goal is to provide insights for future research.

1. Definition of Agricultural Robots

Agricultural robots refer to autonomous equipment designed for agricultural applications, equipped with precise sensing, autonomous decision-making, intelligent control, and automatic execution capabilities. Their core structure consists of four main components: a precise information sensing system, decision-making and control system, operation execution mechanism, and an autonomous mobile platform—often referred to as the "eyes, brain, hands, and feet" of the robot.

• Precise Information Sensing: This component involves multimodal sensing systems based on vision, touch, hearing, and taste technologies to detect spatial environments, target positions, and shapes. Sensing and navigation technologies are central to this, utilizing a combination of sensors like cameras, LiDAR, ultrasound, and GPS to achieve high-precision environmental perception, trajectory planning, and obstacle detection.

• Autonomous Decision-Making and Intelligent Control: This system includes communication, decision-making, and learning components. Data transmission is conducted via multi-machine communication, multi-component communication, production information communication, and decision-making information communication. Technologies like knowledge reasoning, human-machine interaction, and machine learning (including deep learning and reinforcement learning) enable the robot to autonomously identify targets, predict production, make task decisions, and diagnose faults, adjusting its actions based on its experience and environmental feedback.

• Automatic Operation Execution: This section focuses on efficient and robust robotic drives and end-effectors. With technologies like arm design, operation planning, and flexible operation controls, combined with ground-based autonomous mobile platforms, agricultural robots achieve precise targeting, compliant operation, dynamic servo, and effective human-machine interaction. Software and control algorithms allow for multi-task and multi-machine/execution coordination during operations.

In practice, agricultural robots integrate with artificial intelligence, big data, cloud computing, and the Internet of Things (IoT) to form an agricultural robotics application system. These systems are already well-developed and applied in key processes such as fertilization, pest control, yield estimation, grafting, pruning, inspection, harvesting, and transportation in both field and facility planting. Additionally, they are used in livestock farming for feeding, inspection, disinfection, and milking.

2. Global Development of Agricultural Robots

2.1 The United States

In 2011, the National Science Foundation (NSF) of the United States, in collaboration with the Department of Defense (DoD) and the Defense Advanced Research Projects Agency (DARPA), launched the "National Robotics Initiative" (NRI). This initiative aimed to promote research in fundamental robotic science, technology, and systems integration. In 2017, the NRI 2.0 was introduced, investing $30 to $45 million annually to support the development of collaborative robots, focusing on fundamental science, methods, technology, and system integration to promote interactive collaboration across various robot types and scales. In May 2023, the White House released the third edition of the "National Artificial Intelligence R&D Strategic Plan," which identifies the development of more capable and reliable robots as one of its nine key strategies, emphasizing their application in agriculture.

2.2 The European Union

Since 2010, the European Union has steadily increased its investment in robotic research. Through the "Horizon 2020" program, the EU allocated over 700 million euros for robotics research and innovation. The AgROBOfood project, a dedicated agricultural robotics initiative under this program, focuses on developing highly adaptable agricultural robots based on regional production needs. The project also aims to strengthen research and business connections across different regions by establishing a sustainable network of digital innovation hubs. Other projects, such as ROBS4CROPS, CROPS, and TrimBot2020, have been launched to enhance productivity, competitiveness, and sustainable development in agriculture.

2.3 The United Kingdom

The UK has long been a leader in artificial intelligence and robotics, having started research in robotic technology in the 1990s. In 2013, the UK government established the Robotic and Autonomous Systems Interest Group (RAS-SIG) to develop strategic blueprints for robotics and optimize the allocation of resources within the field. In 2014, the UK Technology Strategy Board released the "Robotics and Autonomous Systems 2020 Strategy" (RAS2020), outlining the vision for agricultural robots. This vision focuses on seamless cooperation between robots, farms, and food factories, promoting autonomous collaboration between robots and humans across multiple modes of interaction.

2.4 Japan

Japan is renowned as a "robotics superpower," and in recent years, its need for robotic development and application has grown due to the challenges of an aging and shrinking population. In response, the Japanese government released the "Japan's Robot Strategy" in 2015, identifying agriculture, forestry, fisheries, and the food industry as one of the five key areas for robotic development. The strategy emphasizes expanding the use of robots in environmental sensing, cultivation, harvesting, and livestock wearable technologies, aiming to maintain production stability and reduce labor costs.

Read More

Interview with Yao Na, COO of Tianjin Feima Robotics: Leading Global UAV Innovation and Remote Sensing Solutions

In the Tianjin Free Trade Zone, there is a high-tech enterprise specializing in the research, development, production, and sales of drones and related intelligent hardware and software. This company, Tianjin Feima Robotics Technology Co., Ltd., also provides comprehensive solutions for data acquisition, processing, and dissemination. Yao Na, COO of Feima Robotics, stated that the company is committed to providing customers with one-stop spatial data solutions, focusing on three major areas: aerial surveying and remote sensing, intelligent inspection, and public safety.

1. About Tianjin Feima Robotics Technology Co., Ltd.

Founded in December 2020, Tianjin Feima Robotics is located in the Tianjin Free Trade Zone and serves as a comprehensive base integrating R&D, production, supply chain, operations, services, and training. It is a wholly owned subsidiary of Shenzhen Feima Robotics Co., Ltd., responsible for product development, supply chain management, production, and UAV pilot license training and testing. The company has established standardized facilities, including raw material warehouses, machining workshops, production lines, maintenance workshops, finished product warehouses, reliability labs, engineering labs, comprehensive sensor calibration labs, showrooms, and a UAV training center. With advanced equipment and a complete teaching and training system, it aims to accelerate the integration of production, learning, research, and application in the UAV industry.

“Tianjin Feima Robotics is a key technology development enterprise supported by Tianjin’s government. We began constructing the factory production line on August 4, 2021, and the first drone rolled off the line on January 12, 2022,” Yao Na explained. She further highlighted the company’s strategic products, including the 3D mobile scanning and measurement platforms SLAM100 and SLAM2000, which are globally sold and represent the development of locally born solutions from Tianjin. Currently, the company’s production capacity is stable at 3,000 to 5,000 drones annually, with over 10,000 successful flight operations per year. The company also trains over 2,000 people annually, including 200 under CAAC (Civil Aviation Administration of China) licensure programs.

With the support of various government agencies, Tianjin Feima Robotics has rapidly grown over the past two and a half years, achieving national high-tech enterprise certification and recognition as a technology-based small and medium enterprise. The company has also been certified as a "Hawk" enterprise in Tianjin and has obtained certifications such as DCMM, ISO9001, ISO14001, ISO45001, and ISO27001. Additionally, Tianjin Feima serves as the vice president of both the Tianjin Low-Altitude Economic Alliance and the Tianjin UAV Industry Association, and is a member of the Tianjin UAV and New Materials Alliance.

Feima Robotics has developed a unique series of mass-produced civilian drones, covering fixed-wing, multi-rotor, and vertical takeoff and landing (VTOL) platforms. These 25 drone models range from 3 to 25 kilograms in weight, with payload capacities of 0.5 to 6 kilograms and flight durations of 1 to 4 hours. The drones can support more than 80 types of mission payloads, including orthophotography, oblique photography, LiDAR, multispectral, hyperspectral, dual-light video, and SAR. The company offers integrated hardware and software solutions for spatial data collection across key sectors, such as aerial surveying, intelligent inspections, and public safety, with applications in topographic mapping, land surveys, engineering reconnaissance, smart cities, power grid inspections, railway inspections, water conservation, forest fire prevention, maritime patrols, emergency response, and security surveillance.

2. Core Drone Technologies

“In terms of performance, compared to other drones in the industry, Feima drones stand out in endurance, wind resistance, flight altitude, and adaptability to high-altitude and large-elevation difference environments,” Yao Na explained. The company holds complete proprietary rights to the entire set of drone-related technologies, including aerodynamics, structure, electronics, power, payload, flight control, ground stations, cloud monitoring, and data processing.

Tianjin Feima has also developed an advanced "adaptive UAV flight control system for wide-area complex environments," which addresses key technological challenges such as multi-sensor fusion, real-time collaborative perception, integrated navigation, precise terrain following, wind speed estimation, and fault isolation diagnostics. This ensures stable and intelligent long-duration flight control for aerial surveying and remote sensing missions. The company is also researching real-time ground visualization and cloud monitoring systems, as well as multi-sensor remote image and LiDAR data processing technologies. These efforts help optimize flight route planning and real-time monitoring in complex environments, ensuring safe and accurate aerial data collection.

Moreover, Feima Robotics has contributed to technological innovations in the domestic drone remote sensing and geographic information fields, working alongside other civilian drone manufacturers. This has significantly enhanced the application level and capability of domestically produced drones, accelerating the industrialization of UAV remote sensing equipment and boosting its effectiveness in various sectors.

3. Future Development

The further deepening of reform and opening up in China has provided Tianjin Feima Robotics with an open market and competitive environment, allowing its products and applications to reach internationally advanced levels. Under the “Belt and Road” initiative, the company's drones are now in use across 54 countries and regions, benefiting from global engineering projects that demand high-efficiency, precision mapping, and geospatial data.

In recent years, Feima Robotics has supported China’s Antarctic scientific expeditions, becoming the only domestic UAV manufacturer to provide full-service drone solutions for these polar missions. Its drones have successfully collected large-scale aerial data during several Antarctic expeditions. In 2024, the University of Gothenburg in Sweden purchased two Feima VTOL drones for scientific research in the Amundsen Sea of West Antarctica, marking a significant international milestone for the company's products.

Yao Na emphasized that the company will continue to invest in research and development, focusing on technological innovation and breakthroughs in key performance metrics such as intelligence, efficiency, safety, and reliability. In the future, Tianjin Feima Robotics aims to expand further into international markets and enhance its global competitiveness.

Read More

Research on DGA Malicious Domain Detection Methods

 

As technology advances and the complexity of network environments grows, researchers are increasingly focused on improving the accuracy and efficiency of malicious domain detection to tackle the rising challenges in cybersecurity.

Research on DGA Malicious Domain Detection Methods

1. Blacklist-Based Detection of DGA Domains

In the early days of malicious domain detection, researchers commonly relied on blacklist-based methods to detect domains generated by Domain Generation Algorithms (DGA). Security teams would compile a list of known malicious domains, frequently updating it to ensure it remained current. This list would then be used to block malicious domains or alert users to potential threats. As network technology evolved, these blacklists expanded significantly. Prominent cybersecurity organizations, such as the 360 Network Security Lab, began offering and maintaining publicly accessible domain blacklists. In addition to traditional blacklist approaches, early methods also incorporated DNS flagging techniques to detect suspicious domains.

2. Machine Learning-Based Detection of DGA Domains

Machine learning has proven effective at identifying patterns in large datasets of domain names, significantly improving the detection and classification of malicious domains. As a result, the use of machine learning techniques to detect DGA-based domains has become a prominent focus in this field.

Machine learning approaches generally follow four key steps: data collection, data preprocessing, algorithm development, and evaluation. Early research primarily used static lexical features, which were well-suited for machine learning models. Initial DGA detection methods depended on manually selected domain features, such as domain length, character frequency, and the number of subdomains. As the field progressed, researchers introduced more sophisticated feature extraction techniques, such as entropy-based feature selection and frequency-based feature representation, to improve model accuracy and generalization.

As traditional machine learning techniques evolved, researchers started employing widely recognized algorithms like Naïve Bayes (NB), Random Forest (RF), and Support Vector Machines (SVM) to classify and detect malicious domains.

3. Deep Learning-Based Detection of DGA Domains

In addition to machine learning, deep learning has been increasingly applied to the detection of malicious domains. Traditional detection methods often struggle when faced with complex, high-dimensional domain data, but the emergence of deep learning techniques has provided a more robust solution. Deep learning models have strong feature-learning capabilities, allowing them to automatically extract higher-level, abstract features from large-scale data. This has led to improvements in both the accuracy and reliability of malicious domain detection.

Popular deep learning models used in DGA detection include Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory networks (LSTM), and Generative Adversarial Networks (GAN).

Read More