Biosystems Engineering最新文献_第3页

Toward sustainable urban agriculture: Development of a robotic end effector for hydroponic lettuce in-situ harvesting in vertical farming 面向可持续城市农业：垂直农业中水培莴苣原位收获机器人末端执行器的开发

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-19 DOI: 10.1016/j.biosystemseng.2025.104362

Tiancheng Yu , Huanyu Jiang

Vertical farming integrated with hydroponics offers sustainable urban food solutions, yet reliance on manual harvesting, which requires labour-intensive tray transport and cutting, hinders scalability. This study focuses on the design and performance testing of an end effector for automated in-situ robotic harvesting of vertically stacked hydroponic lettuce, addressing two critical advancements: mechanised harvesting and post-harvest lightweight tray logistics. Manual harvesting trials revealed that push-cutting requires only 3.154 N of force (approximately 66 % lower than pull-cutting's 9.313 N), forming the basis for an energy-efficient end effector design. A novel robotic harvesting mode based on push-cutting was proposed, incorporating five phases: inserting, gripping, cutting, transporting, and unloading. An end effector integrating gripping and cutting functions was designed and optimised. Using response surface methodology, the optimal gripper parameters were determined: rotation radius of 73 mm, rotation angle of 80°, and movable finger diameter of 25 mm. The cutting performance test revealed the optimal cutting position as 5 mm above the root-stem junction and the optimal bevel cutting angle as 15°. Through structural improvements, the end effector was enhanced with a two-finger rotation mechanism to achieve more efficient and convenient unloading. Experimental results from performance testing demonstrated a harvesting success rate of 94 % and a leaf loss rate of 3.50 %, confirming the effectiveness and feasibility of the designed end effector. This innovation establishes a logistics-minimised automation framework for vertical farms, enhancing scalability and sustainability through energy-efficient, high-precision robotics and optimised material flow.

垂直农业与水培法相结合，提供了可持续的城市食品解决方案，但依赖人工收割，这需要劳动密集型的托盘运输和切割，阻碍了可扩展性。本研究的重点是垂直堆叠水培莴苣的自动原位机器人收获的末端执行器的设计和性能测试，解决了两个关键的进步：机械化收获和收获后的轻质托盘物流。人工采集试验表明，推切只需要3.154 N的力（比拉切的9.313 N低约66%），这为节能末端执行器的设计奠定了基础。提出了一种基于推切的机器人采收模式，该模式包括插拔、夹持、切割、运输和卸载五个阶段。设计并优化了集夹持和切削功能于一体的末端执行器。采用响应面法确定了最佳夹持器参数：旋转半径为73 mm，旋转角度为80°，活动指直径为25 mm。切割性能试验表明，最佳切割位置为根茎连接处上方5 mm，最佳斜角为15°。通过结构改进，增强了末端执行器的两指旋转机构，以实现更高效和方便的卸载。性能试验结果表明，采收成功率为94%，叶片损失率为3.50%，验证了所设计末端执行器的有效性和可行性。这一创新为垂直农场建立了物流最小化的自动化框架，通过节能、高精度机器人和优化的物料流增强了可扩展性和可持续性。

{"title":"Toward sustainable urban agriculture: Development of a robotic end effector for hydroponic lettuce in-situ harvesting in vertical farming","authors":"Tiancheng Yu , Huanyu Jiang","doi":"10.1016/j.biosystemseng.2025.104362","DOIUrl":"10.1016/j.biosystemseng.2025.104362","url":null,"abstract":"<div><div>Vertical farming integrated with hydroponics offers sustainable urban food solutions, yet reliance on manual harvesting, which requires labour-intensive tray transport and cutting, hinders scalability. This study focuses on the design and performance testing of an end effector for automated in-situ robotic harvesting of vertically stacked hydroponic lettuce, addressing two critical advancements: mechanised harvesting and post-harvest lightweight tray logistics. Manual harvesting trials revealed that push-cutting requires only 3.154 N of force (approximately 66 % lower than pull-cutting's 9.313 N), forming the basis for an energy-efficient end effector design. A novel robotic harvesting mode based on push-cutting was proposed, incorporating five phases: inserting, gripping, cutting, transporting, and unloading. An end effector integrating gripping and cutting functions was designed and optimised. Using response surface methodology, the optimal gripper parameters were determined: rotation radius of 73 mm, rotation angle of 80°, and movable finger diameter of 25 mm. The cutting performance test revealed the optimal cutting position as 5 mm above the root-stem junction and the optimal bevel cutting angle as 15°. Through structural improvements, the end effector was enhanced with a two-finger rotation mechanism to achieve more efficient and convenient unloading. Experimental results from performance testing demonstrated a harvesting success rate of 94 % and a leaf loss rate of 3.50 %, confirming the effectiveness and feasibility of the designed end effector. This innovation establishes a logistics-minimised automation framework for vertical farms, enhancing scalability and sustainability through energy-efficient, high-precision robotics and optimised material flow.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104362"},"PeriodicalIF":5.3,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Continuous welfare assessment of dairy cows at individual level: A farmer-oriented tool based on normal daily ranges of sensor-recorded traits 奶牛个体水平的连续福利评估：基于传感器记录特征的正常每日范围的面向农民的工具

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-18 DOI: 10.1016/j.biosystemseng.2025.104365

Y. Gómez , N. Blasco-Andreo , A. Llabrés-Brustenga , K. Chow , J. Serra-Sagrista , G.V. Berteselli , E. Canali , X. Manteca , P. Llonch

Animal welfare on farms is currently assessed using human-evaluation protocols, which provide a single record of herd condition at a specific moment. Integrating sensor-based data and farm records, continuous information on each animal's welfare can be obtained. This study aims to create an algorithm to assess individual dairy cow welfare, contributing to the goal of building a platform to inform producers and consumers about dairy cattle welfare. It was built based on the Five Domains model of animal welfare. 221 cows from four commercial free-stall barn farms in Spain and Italy were fitted with accelerometry collars and rumen boluses and monitored for 92 days. Individual data were collected daily. Accelerometers recorded time spent ruminating, eating, lying, walking, and standing within a 24-h interval. Boluses recorded rumen pH and temperature every 10 min, averaged over 24 h. Farm records included parity, veterinary treatments, and milk conductivity. The model provides a daily global welfare index per cow, categorised into health, nutrition, and environment scores. Behaviour and mental state were not included due to a lack of relevant sensor data. Scores range from 0 to 10, indicating the likelihood of the cow experiencing welfare-compromising conditions. Normal thresholds, based on scientific literature, were set for each trait. The algorithm detected daily deviations in traits, assuming that cows with welfare issues deviate from normal behavioural and physiological patterns. When a cow's welfare index decreased, affected domains could be identified, enabling farmers to address potential welfare issues and implement corrective measures.

农场的动物福利目前使用人类评估协议进行评估，该协议提供了特定时刻畜群状况的单一记录。整合基于传感器的数据和农场记录，可以获得关于每只动物福利的连续信息。本研究旨在创建一种评估奶牛个体福利的算法，为建立一个向生产者和消费者通报奶牛福利的平台做出贡献。它是基于动物福利的五域模型建立的。对来自西班牙和意大利4个商业散养牛场的221头奶牛进行了为期92天的监测，并安装了加速计项圈和瘤胃丸。每天收集个人数据。加速度计记录了24小时间隔内反刍、进食、躺卧、行走和站立的时间。boles每10分钟记录一次瘤胃pH和温度，平均记录时间为24小时。农场记录包括胎次、兽医处理和乳传导率。该模型提供每头奶牛的每日全球福利指数，分为健康、营养和环境得分。由于缺乏相关的传感器数据，行为和精神状态未包括在内。得分范围从0到10，表示奶牛遭受福利损害的可能性。在科学文献的基础上，为每个特征设定了正常阈值。该算法检测到性状的日常偏差，假设有福利问题的奶牛偏离了正常的行为和生理模式。当奶牛的福利指数下降时，可以识别受影响的领域，使农民能够解决潜在的福利问题并实施纠正措施。

{"title":"Continuous welfare assessment of dairy cows at individual level: A farmer-oriented tool based on normal daily ranges of sensor-recorded traits","authors":"Y. Gómez , N. Blasco-Andreo , A. Llabrés-Brustenga , K. Chow , J. Serra-Sagrista , G.V. Berteselli , E. Canali , X. Manteca , P. Llonch","doi":"10.1016/j.biosystemseng.2025.104365","DOIUrl":"10.1016/j.biosystemseng.2025.104365","url":null,"abstract":"<div><div>Animal welfare on farms is currently assessed using human-evaluation protocols, which provide a single record of herd condition at a specific moment. Integrating sensor-based data and farm records, continuous information on each animal's welfare can be obtained. This study aims to create an algorithm to assess individual dairy cow welfare, contributing to the goal of building a platform to inform producers and consumers about dairy cattle welfare. It was built based on the Five Domains model of animal welfare. 221 cows from four commercial free-stall barn farms in Spain and Italy were fitted with accelerometry collars and rumen boluses and monitored for 92 days. Individual data were collected daily. Accelerometers recorded time spent ruminating, eating, lying, walking, and standing within a 24-h interval. Boluses recorded rumen pH and temperature every 10 min, averaged over 24 h. Farm records included parity, veterinary treatments, and milk conductivity. The model provides a daily global welfare index per cow, categorised into health, nutrition, and environment scores. Behaviour and mental state were not included due to a lack of relevant sensor data. Scores range from 0 to 10, indicating the likelihood of the cow experiencing welfare-compromising conditions. Normal thresholds, based on scientific literature, were set for each trait. The algorithm detected daily deviations in traits, assuming that cows with welfare issues deviate from normal behavioural and physiological patterns. When a cow's welfare index decreased, affected domains could be identified, enabling farmers to address potential welfare issues and implement corrective measures.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104365"},"PeriodicalIF":5.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788840","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Interaction mechanisms of seed suction and cleaning negative pressure of an air-suction seed metering device for maize breeding plots 玉米养殖场空气吸种器吸种与清洗负压的相互作用机理

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-18 DOI: 10.1016/j.biosystemseng.2025.104368

Pengfei Zhao , Qingbin Song , Yubin Bi , Jianxin Dong , Zuoli Fu , Yuxiang Huang , Yi Zheng

To study the interaction mechanism between the seed suction and cleaning negative pressure at the suction hole of the air-suction seed metering device in maize breeding plot, CFD was used to simulate the flow field at the suction hole during the seed cleaning process. The decay change law of pressure at the suction hole during seed cleaning was studied through single-factor tests. The results showed that the adsorption pressure decay rate of the suction holes in the seed filling area was the largest. By analysing the pressure field and velocity field at the suction hole, it was found that the minimum negative pressure centre of the suction hole was affected by the seed cleaning negative pressure and shifted to the seed cleaning airflow side. Through the central composite design test, it was found that the seed suction and cleaning negative pressure had a highly significant interactive effect on the adsorption pressure of the suction holes in the seed filling area. And the influence of seed suction negative pressure was greater than that of seed cleaning negative pressure. The mathematical models for predicting adsorption pressure and decay rate were established on this basis. Through bench tests, the optimal parameter combination was determined to be 7 kPa for seed suction and 5 kPa for seed cleaning. No dropping of adsorbed seeds in the seed filling area was basically realized. This study can provide a theoretical basis for the design of air-suction seed metering devices in the plots.

为研究玉米养殖场空气吸收式排种装置吸种孔处吸种与清洗负压的相互作用机理，采用CFD模拟了玉米养殖场吸种孔清洗过程中的流场。通过单因素试验，研究了种子清洗过程中吸气孔压力的衰减变化规律。结果表明：灌种区吸气孔的吸附压力衰减率最大；通过分析吸气孔处的压力场和速度场，发现吸气孔的最小负压中心受到洗种负压的影响，向洗种气流侧偏移。通过中心复合设计试验，发现种子吸力和清洗负压对种子填充区吸力孔的吸附压力具有高度显著的交互作用。吸种负压的影响大于洗种负压的影响。在此基础上建立了预测吸附压力和衰减率的数学模型。通过台架试验，确定最优参数组合为吸籽7 kPa，清籽5 kPa。基本实现了灌种区吸附种子不滴落。该研究可为小区内空气吸式排种装置的设计提供理论依据。

{"title":"Interaction mechanisms of seed suction and cleaning negative pressure of an air-suction seed metering device for maize breeding plots","authors":"Pengfei Zhao , Qingbin Song , Yubin Bi , Jianxin Dong , Zuoli Fu , Yuxiang Huang , Yi Zheng","doi":"10.1016/j.biosystemseng.2025.104368","DOIUrl":"10.1016/j.biosystemseng.2025.104368","url":null,"abstract":"<div><div>To study the interaction mechanism between the seed suction and cleaning negative pressure at the suction hole of the air-suction seed metering device in maize breeding plot, CFD was used to simulate the flow field at the suction hole during the seed cleaning process. The decay change law of pressure at the suction hole during seed cleaning was studied through single-factor tests. The results showed that the adsorption pressure decay rate of the suction holes in the seed filling area was the largest. By analysing the pressure field and velocity field at the suction hole, it was found that the minimum negative pressure centre of the suction hole was affected by the seed cleaning negative pressure and shifted to the seed cleaning airflow side. Through the central composite design test, it was found that the seed suction and cleaning negative pressure had a highly significant interactive effect on the adsorption pressure of the suction holes in the seed filling area. And the influence of seed suction negative pressure was greater than that of seed cleaning negative pressure. The mathematical models for predicting adsorption pressure and decay rate were established on this basis. Through bench tests, the optimal parameter combination was determined to be 7 kPa for seed suction and 5 kPa for seed cleaning. No dropping of adsorbed seeds in the seed filling area was basically realized. This study can provide a theoretical basis for the design of air-suction seed metering devices in the plots.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104368"},"PeriodicalIF":5.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788842","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Design and optimisation of an air-assisted helical seed-guiding device for high-speed maize planters using DEM-CFD coupling 基于DEM-CFD耦合的高速玉米播种机气动螺旋导种装置设计与优化

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-18 DOI: 10.1016/j.biosystemseng.2025.104356

Long Zhou , Kuo Sun , Xu Zhao , Jingxiang Wang , Yulong Chen , Yongchang Sun , Wenjun Wang

Seeds are easy to collide with the guiding tube during air-seed contact in the air-assisted seed guiding process, which seriously affects the quality of sowing and has become a bottleneck problem hindering the development of precision sowing technology. Responding to the above, an innovative design of helical seed-guiding device with pressurised airflow assistance is presented. The structural parameters of the device are then optimised using Discrete Element Method - Computational Fluid Dynamics (DEM- CFD) simulation. To solve the problem of low accuracy of the ‘bonded-sphere model + unresolved DEM-CFD coupling model’, the calculation of porosity is improved by using virtual dual-grid porosity model with Gaussian kernel function smoothing method, and then the ‘multi-sphere model + semi-resolved DEM-CFD coupling model’ is constructed. Comparison of tests using blowing seeds with positive-pressure airflow reveals that the latter improves the simulation accuracy by 42.79 % relative to the former. Based on the new coupling calculation method, significance analysis and optimisation of the structural parameters of the helical seed-guiding device with pressurised airflow assistance by combining the Plackett-Burman and Box-Behnken test. The optimum combination of parameters obtained from the optimisation is a width of 4 mm for the helical guided airflow chamber, an angle of 39.36° for the inlet pipe and a number of helixes of 3. Finally, the working performance of the optimised seed guide tube is validated by adaptive and bench tests. The results showed that the coefficients of variation of plant spacing at each operating speed (8 km⋅h⁻¹, 10 km⋅h⁻¹, 12 km⋅h⁻¹, and 14 km⋅h⁻¹) are 5.51 %, 6.09 %, 9.01 %, 9.30 %, respectively. This study not only proposes a new airflow-assisted seed guiding device but also provides a coupled DEM-CFD calculation method applicable to large-size non-spherical particles.

在气助导种过程中，种子在接触过程中容易与导种管发生碰撞，严重影响播种质量，成为阻碍精准播种技术发展的瓶颈问题。针对上述问题，提出了一种具有增压气流辅助的螺旋导种装置的创新设计。然后使用离散元法-计算流体动力学（DEM- CFD）模拟优化设备的结构参数。针对“粘接球模型+未解析DEM-CFD耦合模型”精度低的问题，采用高斯核函数平滑法对孔隙度的虚拟双网格孔隙度模型进行改进，构建了“多球模型+半解析DEM-CFD耦合模型”。与正压吹籽试验对比表明，正压吹籽比正压吹籽提高了42.79%的模拟精度。基于新的耦合计算方法，结合Plackett-Burman和Box-Behnken试验，对增压气流辅助下螺旋导种装置结构参数进行了显著性分析和优化。优化得到的最佳参数组合为螺旋引导气流室宽度为4mm，进气管角度为39.36°，螺旋数为3。最后，通过自适应试验和台架试验验证了优化后的导种管的工作性能。结果表明：8 km⋅h−1、10 km⋅h−1、12 km⋅h−1、14 km⋅h−1各运行速度下的株距变异系数分别为5.51%、6.09%、9.01%、9.30%；本研究不仅提出了一种新的气流辅助导粒装置，而且提供了一种适用于大粒径非球形颗粒的DEM-CFD耦合计算方法。

{"title":"Design and optimisation of an air-assisted helical seed-guiding device for high-speed maize planters using DEM-CFD coupling","authors":"Long Zhou , Kuo Sun , Xu Zhao , Jingxiang Wang , Yulong Chen , Yongchang Sun , Wenjun Wang","doi":"10.1016/j.biosystemseng.2025.104356","DOIUrl":"10.1016/j.biosystemseng.2025.104356","url":null,"abstract":"<div><div>Seeds are easy to collide with the guiding tube during air-seed contact in the air-assisted seed guiding process, which seriously affects the quality of sowing and has become a bottleneck problem hindering the development of precision sowing technology. Responding to the above, an innovative design of helical seed-guiding device with pressurised airflow assistance is presented. The structural parameters of the device are then optimised using Discrete Element Method - Computational Fluid Dynamics (DEM- CFD) simulation. To solve the problem of low accuracy of the ‘bonded-sphere model + unresolved DEM-CFD coupling model’, the calculation of porosity is improved by using virtual dual-grid porosity model with Gaussian kernel function smoothing method, and then the ‘multi-sphere model + semi-resolved DEM-CFD coupling model’ is constructed. Comparison of tests using blowing seeds with positive-pressure airflow reveals that the latter improves the simulation accuracy by 42.79 % relative to the former. Based on the new coupling calculation method, significance analysis and optimisation of the structural parameters of the helical seed-guiding device with pressurised airflow assistance by combining the Plackett-Burman and Box-Behnken test. The optimum combination of parameters obtained from the optimisation is a width of 4 mm for the helical guided airflow chamber, an angle of 39.36° for the inlet pipe and a number of helixes of 3. Finally, the working performance of the optimised seed guide tube is validated by adaptive and bench tests. The results showed that the coefficients of variation of plant spacing at each operating speed (8 km⋅h<sup>−1</sup>, 10 km⋅h<sup>−1</sup>, 12 km⋅h<sup>−1</sup>, and 14 km⋅h<sup>−1</sup>) are 5.51 %, 6.09 %, 9.01 %, 9.30 %, respectively. This study not only proposes a new airflow-assisted seed guiding device but also provides a coupled DEM-CFD calculation method applicable to large-size non-spherical particles.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104356"},"PeriodicalIF":5.3,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A tractor tyre rut depth model based on DEM-MBD coupling simulation for improved tillage depth measurement 基于DEM-MBD耦合仿真的拖拉机轮胎车辙深度模型改进耕作深度测量

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-16 DOI: 10.1016/j.biosystemseng.2025.104361

Xiaoxiang Weng, Xiaobo Xi, Yangjie Shi, Ruihong Zhang

Real-time tillage depth information is crucial for ensuring normal seed development. However, existing measurement devices often overlook the effect of tyre rut depth on measurement errors. Taking the alluvial soil in the rice-wheat rotation area of Yangzhou, China as the study area, parameters of the Edinburgh Elasto-Plastic Adhesion (EEPA) contact model were calibrated through uniaxial compression experiments, and a soil-straw discrete element soil bin model was established. The Discrete Element Method-Multibody Dynamics (DEM-MBD) coupled simulation was employed to analyse the effects of forward speed, soil penetration resistance, and straw mulching amount on tyre rut depth. The results showed that rut depth decreased by 5.2 mm as forward speed increased from 1.5 to 7.5 km h⁻¹, by 43.8 mm as soil penetration resistance increased from 1 to 3 MPa, and by 3.9 mm as straw mulching amount increased from 0 to 1.2 kg m⁻². These results indicated that soil penetration resistance is the dominant factor affecting rut depth, while forward speed and straw mulching amount have relatively minor effects. A mathematical model was established to quantitatively predict rut depth based on soil penetration resistance, and its reliability was validated through field compaction experiments. Incorporating the rut depth model into the tillage depth measurement system significantly reduced measurement errors. Field experiments showed that higher forward speeds led to increased measurement errors and reduced operation quality. An optimal forward speed of 4.5 km h⁻¹ was recommended, at which the tillage depth measurement error was ±8.2 mm, meeting the requirements for tillage depth measurement.

实时耕作深度信息对确保种子正常发育至关重要。然而，现有的测量装置往往忽略了轮胎车辙深度对测量误差的影响。以扬州稻麦轮作区冲积土为研究区，通过单轴压缩试验对爱丁堡弹塑性粘着（EEPA）接触模型参数进行了标定，建立了土壤-秸秆离散元土仓模型。采用离散元法-多体动力学（DEM-MBD）耦合仿真分析了前进速度、土壤穿透阻力和秸秆覆盖量对轮胎车辙深度的影响。结果表明：当前进速度从1.5 km h−1增加到7.5 km h−1时，车辙深度减少5.2 mm；土壤穿透阻力从1 MPa增加到3 MPa时，车辙深度减少43.8 mm；秸秆覆盖量从0 kg m−2增加到1.2 kg m−2时，车辙深度减少3.9 mm。结果表明，土壤穿透阻力是影响车辙深度的主要因素，而前进速度和秸秆覆盖量的影响相对较小。建立了基于土壤穿透阻力定量预测车辙深度的数学模型，并通过现场压实试验验证了模型的可靠性。将车辙深度模型纳入耕深测量系统，显著降低了测量误差。现场实验表明，较高的前向速度会增加测量误差，降低操作质量。建议最佳前进速度为4.5 km h−1，在此速度下，耕深测量误差为±8.2 mm，满足耕深测量要求。

{"title":"A tractor tyre rut depth model based on DEM-MBD coupling simulation for improved tillage depth measurement","authors":"Xiaoxiang Weng, Xiaobo Xi, Yangjie Shi, Ruihong Zhang","doi":"10.1016/j.biosystemseng.2025.104361","DOIUrl":"10.1016/j.biosystemseng.2025.104361","url":null,"abstract":"<div><div>Real-time tillage depth information is crucial for ensuring normal seed development. However, existing measurement devices often overlook the effect of tyre rut depth on measurement errors. Taking the alluvial soil in the rice-wheat rotation area of Yangzhou, China as the study area, parameters of the Edinburgh Elasto-Plastic Adhesion (EEPA) contact model were calibrated through uniaxial compression experiments, and a soil-straw discrete element soil bin model was established. The Discrete Element Method-Multibody Dynamics (DEM-MBD) coupled simulation was employed to analyse the effects of forward speed, soil penetration resistance, and straw mulching amount on tyre rut depth. The results showed that rut depth decreased by 5.2 mm as forward speed increased from 1.5 to 7.5 km h<sup>−1</sup>, by 43.8 mm as soil penetration resistance increased from 1 to 3 MPa, and by 3.9 mm as straw mulching amount increased from 0 to 1.2 kg m<sup>−2</sup>. These results indicated that soil penetration resistance is the dominant factor affecting rut depth, while forward speed and straw mulching amount have relatively minor effects. A mathematical model was established to quantitatively predict rut depth based on soil penetration resistance, and its reliability was validated through field compaction experiments. Incorporating the rut depth model into the tillage depth measurement system significantly reduced measurement errors. Field experiments showed that higher forward speeds led to increased measurement errors and reduced operation quality. An optimal forward speed of 4.5 km h<sup>−1</sup> was recommended, at which the tillage depth measurement error was ±8.2 mm, meeting the requirements for tillage depth measurement.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104361"},"PeriodicalIF":5.3,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimising refrigerated container cooling performance: A virtual modelling approach for temperature and quality management in fresh fruit cold chain 优化冷藏容器冷却性能：在新鲜水果冷链温度和质量管理的虚拟建模方法

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-16 DOI: 10.1016/j.biosystemseng.2025.104357

Chijioke Leonard Nkwocha , Alemayehu Ambaw Tsige , Buhle Maphosa , Corné Coetzee , Umezuruike Linus Opara

Effective temperature management in refrigerated containers (reefers) is essential for preserving fruit quality during long-distance cold chain transport. This study employs a virtual modelling approach to assess and optimise reefer cooling performance, focusing on spatial and temporal temperature variations within a fully loaded container transporting fresh fruit. Time-series data was obtained from embedded sensors which monitored fruit core and ambient air temperatures at six locations: front left (FL), front right (FR), middle left (ML), middle right (MR), back left (BL), and back right (BR). Results revealed pronounced thermal asymmetry, with the left side, especially FL and ML, exhibiting the slowest cooling response, remaining above 1 °C for nearly 190 h, compared to under 130 h for MR and BL. This disparity, only partly attributable to minor differences in pallet airflow resistance, may be influenced by external heat gain, potentially due to solar exposure depending on container orientation during maritime transport. Crucially, the study recommends using fruit-core temperature from a pallet near the container door as a key indicator of residual cooling needs and pre-optimality prior to shipment. It also highlights the importance of accounting for apparent ambient temperature and time–temperature exposure before vessel departure in determining proper loading conditions. Further experimental work is encouraged to investigate the influence of container placement on the ship deck and its interaction with external environmental conditions.

在长途冷链运输中，有效的冷藏容器（冷藏箱）温度管理对于保持水果品质至关重要。本研究采用虚拟建模方法来评估和优化冷藏箱冷却性能，重点关注运输新鲜水果的满载集装箱内的空间和时间温度变化。时间序列数据来自嵌入式传感器，这些传感器监测果核和六个位置的环境空气温度：前左（FL）、前右（FR）、中左（ML）、中右（MR）、后左（BL）和后右（BR）。结果显示了明显的热不对称，左侧，特别是FL和ML，表现出最慢的冷却响应，保持在1°C以上近190小时，而MR和BL则低于130小时。这种差异仅部分归因于托盘气流阻力的微小差异，可能受到外部热增益的影响，可能是由于海上运输过程中集装箱朝向的太阳照射。至关重要的是，该研究建议使用集装箱门附近托盘的果核温度作为运输前剩余冷却需求和预优化的关键指标。它还强调了在船舶出发前考虑表观环境温度和时间温度暴露的重要性，以确定适当的装载条件。鼓励进一步的实验工作来研究集装箱放置对船舶甲板的影响及其与外部环境条件的相互作用。

{"title":"Optimising refrigerated container cooling performance: A virtual modelling approach for temperature and quality management in fresh fruit cold chain","authors":"Chijioke Leonard Nkwocha , Alemayehu Ambaw Tsige , Buhle Maphosa , Corné Coetzee , Umezuruike Linus Opara","doi":"10.1016/j.biosystemseng.2025.104357","DOIUrl":"10.1016/j.biosystemseng.2025.104357","url":null,"abstract":"<div><div>Effective temperature management in refrigerated containers (reefers) is essential for preserving fruit quality during long-distance cold chain transport. This study employs a virtual modelling approach to assess and optimise reefer cooling performance, focusing on spatial and temporal temperature variations within a fully loaded container transporting fresh fruit. Time-series data was obtained from embedded sensors which monitored fruit core and ambient air temperatures at six locations: front left (FL), front right (FR), middle left (ML), middle right (MR), back left (BL), and back right (BR). Results revealed pronounced thermal asymmetry, with the left side, especially FL and ML, exhibiting the slowest cooling response, remaining above 1 °C for nearly 190 h, compared to under 130 h for MR and BL. This disparity, only partly attributable to minor differences in pallet airflow resistance, may be influenced by external heat gain, potentially due to solar exposure depending on container orientation during maritime transport. Crucially, the study recommends using fruit-core temperature from a pallet near the container door as a key indicator of residual cooling needs and pre-optimality prior to shipment. It also highlights the importance of accounting for apparent ambient temperature and time–temperature exposure before vessel departure in determining proper loading conditions. Further experimental work is encouraged to investigate the influence of container placement on the ship deck and its interaction with external environmental conditions.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104357"},"PeriodicalIF":5.3,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788836","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing trajectory overlap of follower robots via a novel circular constraint following paradigm in human-robot collaborative fruit harvesting 在人-机器人协同水果收获中，利用一种新的圆形约束跟随范式增强跟随机器人的轨迹重叠

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-15 DOI: 10.1016/j.biosystemseng.2025.104359

Hengda Li, Liang Sun, Zhenghao Li, Ying Chen, Haichao Li, Pingyi Liu

To enhance the efficiency and automation of collaborative fruit harvesting, this study investigates follower robots operating in greenhouse environments. It identifies the trajectory offset problem inherent in the traditional directional constraint following paradigm (DCFP), which limits its applicability in confined spaces. Through theoretical analysis, simulation, and experimental validation, the study reveals the underlying mechanisms of trajectory offset, showing that the deviation increases with both the predefined following distance and the curvature of the human path. To address this issue, a novel following paradigm, the circular constraint following paradigm (CCFP), is proposed along with a detailed implementation method. Field experimental results in a greenhouse peach orchard demonstrate that CCFP achieves lower trajectory offset than DCFP, as quantified by a 32.9 % reduction in average matching cost (AMC) across four typical paths. This indicates that CCFP offers a higher trajectory overlap, enabling collision-free navigation with only the prior path provided by the picker, thus avoiding the need for complex obstacle avoidance systems. Additionally, CCFP is compatible with existing positioning and control models, offering a modular and scalable framework for the future development of agricultural following robots in constrained environments.

为了提高协同水果收获的效率和自动化程度，本研究研究了在温室环境下操作的跟随机器人。它识别了传统定向约束跟随范式（DCFP）固有的轨迹偏移问题，这限制了其在受限空间中的适用性。通过理论分析、仿真和实验验证，揭示了轨迹偏移的潜在机制，表明轨迹偏移随预定跟随距离和人类路径曲率的增加而增加。为了解决这一问题，本文提出了一种新的跟随范式，即循环约束跟随范式（CCFP），并给出了详细的实现方法。温室桃园的田间试验结果表明，CCFP比DCFP实现了更低的轨迹偏移，在四个典型路径上的平均匹配成本（AMC）降低了32.9%。这表明CCFP提供了更高的轨迹重叠，仅使用拾取器提供的先验路径就可以实现无碰撞导航，从而避免了对复杂避障系统的需求。此外，CCFP与现有的定位和控制模型兼容，为限制环境下农业跟随机器人的未来发展提供了模块化和可扩展的框架。

{"title":"Enhancing trajectory overlap of follower robots via a novel circular constraint following paradigm in human-robot collaborative fruit harvesting","authors":"Hengda Li, Liang Sun, Zhenghao Li, Ying Chen, Haichao Li, Pingyi Liu","doi":"10.1016/j.biosystemseng.2025.104359","DOIUrl":"10.1016/j.biosystemseng.2025.104359","url":null,"abstract":"<div><div>To enhance the efficiency and automation of collaborative fruit harvesting, this study investigates follower robots operating in greenhouse environments. It identifies the trajectory offset problem inherent in the traditional directional constraint following paradigm (DCFP), which limits its applicability in confined spaces. Through theoretical analysis, simulation, and experimental validation, the study reveals the underlying mechanisms of trajectory offset, showing that the deviation increases with both the predefined following distance and the curvature of the human path. To address this issue, a novel following paradigm, the circular constraint following paradigm (CCFP), is proposed along with a detailed implementation method. Field experimental results in a greenhouse peach orchard demonstrate that CCFP achieves lower trajectory offset than DCFP, as quantified by a 32.9 % reduction in average matching cost (AMC) across four typical paths. This indicates that CCFP offers a higher trajectory overlap, enabling collision-free navigation with only the prior path provided by the picker, thus avoiding the need for complex obstacle avoidance systems. Additionally, CCFP is compatible with existing positioning and control models, offering a modular and scalable framework for the future development of agricultural following robots in constrained environments.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104359"},"PeriodicalIF":5.3,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788837","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Quantitative alignment and study of low-frequency excitation parameters based on walnut tree (Juglans regia L.) shape characteristics 基于核桃树形状特征的低频激励参数定量比对与研究

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-11 DOI: 10.1016/j.biosystemseng.2025.104355

Jiale Cao , Heng Chen , Jianshuo An , Jingwei Wang , Latai Ga , Daochun Xu , Xiaopeng Bai , Wenbin Li

Walnut trees vary greatly in shape, which results in large differences in the optimal excitation parameters between trees. By clarifying the correlation between the tree shape characteristic parameters and the optimal excitation parameters, the excitation parameters can be matched according to the tree shape characteristics. The purpose of this is to improve picking efficiency and simultaneously reduce damage to the tree. In this study, a field experiment was conducted using a low-frequency shaker (Trunk shakers) and the vibration response under different excitation parameters was analysed using statistical methods. Finally, the range of excitation parameters based on the quantitative matching of tree feature parameters was obtained through linear regression and a comparison of feature importance. It was found that the tree height and crown projection area were the two most important factors affecting the acceleration of the tree vibration response, and they were significantly negatively correlated with each other. The optimal intervals of the excitation frequency and amplitude were successfully predicted from the tree height and canopy projection area as (ω₀η_min, ω₀η_max) and (A₀η_min/β(ω_min), A_maxη_max/β(ω_max)), respectively. Validated tests showed a prediction accuracy of 93 % for this range. This study provides a reliable range of excitation parameters for future walnut harvesting operations, which can guide the development of low-frequency intelligent shakers, in addition to field operations.

核桃树的形状差异很大，这导致不同树间的最优激励参数差异很大。通过理清树形特征参数与最优激励参数之间的相关性，可以根据树形特征匹配激励参数。这样做的目的是提高采摘效率，同时减少对树木的伤害。本研究采用低频振动筛（Trunk shakers）进行了现场试验，采用统计方法分析了不同激励参数下的振动响应。最后，通过线性回归和特征重要度比较，得到基于树形特征参数定量匹配的激励参数范围。结果表明，树高和树冠投影面积是影响树木振动响应加速的两个最重要因素，且两者之间呈显著负相关。由树高和树冠投影面积分别预测出激励频率和振幅的最佳区间为（ω ω ηmin, ω ω ηmax）和(ω ω ηmin/β(ωmin), Amaxηmax/β(ωmax))。经过验证的测试表明，该范围的预测精度为93%。该研究为未来核桃收获作业提供了可靠的激励参数范围，可以指导低频智能激振器的开发，以及现场作业。

{"title":"Quantitative alignment and study of low-frequency excitation parameters based on walnut tree (Juglans regia L.) shape characteristics","authors":"Jiale Cao , Heng Chen , Jianshuo An , Jingwei Wang , Latai Ga , Daochun Xu , Xiaopeng Bai , Wenbin Li","doi":"10.1016/j.biosystemseng.2025.104355","DOIUrl":"10.1016/j.biosystemseng.2025.104355","url":null,"abstract":"<div><div>Walnut trees vary greatly in shape, which results in large differences in the optimal excitation parameters between trees. By clarifying the correlation between the tree shape characteristic parameters and the optimal excitation parameters, the excitation parameters can be matched according to the tree shape characteristics. The purpose of this is to improve picking efficiency and simultaneously reduce damage to the tree. In this study, a field experiment was conducted using a low-frequency shaker (Trunk shakers) and the vibration response under different excitation parameters was analysed using statistical methods. Finally, the range of excitation parameters based on the quantitative matching of tree feature parameters was obtained through linear regression and a comparison of feature importance. It was found that the tree height and crown projection area were the two most important factors affecting the acceleration of the tree vibration response, and they were significantly negatively correlated with each other. The optimal intervals of the excitation frequency and amplitude were successfully predicted from the tree height and canopy projection area as (<em>ω</em><sub>0</sub><em>η</em><sub>min</sub>, <em>ω</em><sub>0</sub><em>η</em><sub>max</sub>) and (<em>A</em><sub>0</sub><em>η</em><sub>min</sub>/<em>β</em>(<em>ω</em><sub>min</sub>), <em>A</em><sub>max</sub><em>η</em><sub>max</sub>/<em>β</em>(<em>ω</em><sub>max</sub>)), respectively. Validated tests showed a prediction accuracy of 93 % for this range. This study provides a reliable range of excitation parameters for future walnut harvesting operations, which can guide the development of low-frequency intelligent shakers, in addition to field operations.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104355"},"PeriodicalIF":5.3,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145712312","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Research on a vision navigation system for rice transplanters based on machine vision fused with GNSS 基于机器视觉与GNSS融合的插秧机视觉导航系统研究

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-12-11 DOI: 10.1016/j.biosystemseng.2025.104354

Zhaopeng Liu , Xulong Wu , Weiwei Gao , Fengpeng Ning , Wenyu Zhang , Jie He , Muhua Liu , Xiongfei Chen , Peng Fang

To achieve the high-precision and continuous autonomous operation of rice transplanters in complex paddy field environments, this study proposes a visual navigation system based on machine vision and GNSS. First, using the Jetson AGX Xavier platform, the BiSeNet multi-class semantic segmentation model identified side and headland ridges. A three-dimensional dynamic inverse perspective transformation and a homogeneous coordinate transformation with integrated 3D camera pose data and GNSS positioning were used to extract the navigation line coordinates of paddy ridges for machine guidance. Subsequently, the side ridge coordinates were offset to construct the first-row dynamic path planning unit, enabling the real-time tracking of ridge boundaries, while the headland ridge distance threshold triggered autonomous steering and row-shifting path planning units, allowing the rice transplanter to make decisions for continuous autonomous operation in complex paddy fields. The navigation controller was developed through secondary development, using a custom communication protocol, integrating hardware and software for the rice transplanter's autonomous operation. Finally, field tests were conducted at a speed of 0.7 m s⁻¹, demonstrating the rice transplanter's ability to autonomously travel along ridge boundaries and shift rows for continuous operation. The experimental results showed that the visual algorithm took approximately 100 ms, with an average error of 0.071 m in navigation line extraction, 0.089 m in headland distance detection, and 0.061 m in path tracking. The results indicated that the proposed visual navigation system enabled the continuous autonomous operation of the rice transplanter and thus provide significant technical support for the development of agricultural robots.

为了实现水稻插秧机在复杂水田环境下的高精度连续自主操作，本研究提出了一种基于机器视觉和GNSS的视觉导航系统。首先，利用Jetson AGX Xavier平台，BiSeNet多类语义分割模型识别了侧脊和海陆脊；利用三维动态反透视变换和三维相机位姿数据与GNSS定位相结合的齐次坐标变换，提取水稻垄导航线坐标，用于机器制导。随后，偏移侧脊坐标，构建第一行动态路径规划单元，实现对脊边界的实时跟踪，而海陆脊距离阈值触发自主转向和行移路径规划单元，使插秧机能够在复杂水田中进行连续自主作业决策。导航控制器通过二次开发，采用自定义通信协议，将硬件和软件集成在一起，实现插秧机的自主操作。最后，以0.7 m s - 1的速度进行了田间试验，证明了水稻插秧机能够自主沿着山脊边界移动并连续移动行。实验结果表明，视觉算法耗时约100 ms，导航线提取平均误差为0.071 m，海岬距离检测平均误差为0.089 m，路径跟踪平均误差为0.061 m。结果表明，该视觉导航系统实现了水稻插秧机的连续自主操作，为农业机器人的发展提供了重要的技术支持。

{"title":"Research on a vision navigation system for rice transplanters based on machine vision fused with GNSS","authors":"Zhaopeng Liu , Xulong Wu , Weiwei Gao , Fengpeng Ning , Wenyu Zhang , Jie He , Muhua Liu , Xiongfei Chen , Peng Fang","doi":"10.1016/j.biosystemseng.2025.104354","DOIUrl":"10.1016/j.biosystemseng.2025.104354","url":null,"abstract":"<div><div>To achieve the high-precision and continuous autonomous operation of rice transplanters in complex paddy field environments, this study proposes a visual navigation system based on machine vision and GNSS. First, using the Jetson AGX Xavier platform, the BiSeNet multi-class semantic segmentation model identified side and headland ridges. A three-dimensional dynamic inverse perspective transformation and a homogeneous coordinate transformation with integrated 3D camera pose data and GNSS positioning were used to extract the navigation line coordinates of paddy ridges for machine guidance. Subsequently, the side ridge coordinates were offset to construct the first-row dynamic path planning unit, enabling the real-time tracking of ridge boundaries, while the headland ridge distance threshold triggered autonomous steering and row-shifting path planning units, allowing the rice transplanter to make decisions for continuous autonomous operation in complex paddy fields. The navigation controller was developed through secondary development, using a custom communication protocol, integrating hardware and software for the rice transplanter's autonomous operation. Finally, field tests were conducted at a speed of 0.7 m s<sup>−1</sup>, demonstrating the rice transplanter's ability to autonomously travel along ridge boundaries and shift rows for continuous operation. The experimental results showed that the visual algorithm took approximately 100 ms, with an average error of 0.071 m in navigation line extraction, 0.089 m in headland distance detection, and 0.061 m in path tracking. The results indicated that the proposed visual navigation system enabled the continuous autonomous operation of the rice transplanter and thus provide significant technical support for the development of agricultural robots.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"262 ","pages":"Article 104354"},"PeriodicalIF":5.3,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145735835","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Aggression recognition for individual pigs based on YOLOv8 and DCTM-UniformerV2 基于YOLOv8和DCTM-UniformerV2的猪个体攻击识别

IF 5.3 1区农林科学 Q1 AGRICULTURAL ENGINEERING

Biosystems Engineering

Pub Date : 2025-11-29 DOI: 10.1016/j.biosystemseng.2025.104345

Chen Chen , Rui Liu , Weixing Zhu , Tomas Norton

The aim of this study is to develop a deep learning algorithm based on YOLOv8 and DCTM-UniformerV2 to recognise aggressive behaviour in pigs. Compared with previous studies, this study further investigates aggression recognition from individual perspectives. In the experiment, 8 nursery pigs were mixed for 3 days and then 8 h of video was recorded each day. 410 aggressive 2 s episodes were labelled from the first day of video. By segmenting each pig in these episodes, sub-episodes of 820 aggressive and 2460 non-aggressive pigs were generated. These 820 aggressive sub-episodes were then augmented into 2460 sub-episodes by using horizontal and vertical mirroring. All the 4920 sub-episodes were divided into the training and validation sets at a ratio of 7:3. Moreover, 190 aggressive 2 s episodes were labelled from the other 2 days of videos, and then 380 aggressive and 1140 non-aggressive sub-episodes were generated as the test set. Firstly, identification module of YOLOv8 was used to locate each pig and determine their identities. Secondly, temporal shift module and depth channel attention mechanism were fused into the aggression recognition module of UniformerV2 to improve the discrimination of spatial-temporal features. Finally, the improved DCTM-UniformerV2 network was used to recognise aggressive behaviour of pigs. In the test set, the aggression could be recognised by using the proposed algorithm with an accuracy of 97.2 %. The result indicates that the proposed method can be used to recognise aggressive behaviour of individual pigs.

本研究的目的是开发一种基于YOLOv8和DCTM-UniformerV2的深度学习算法来识别猪的攻击行为。与以往的研究相比，本研究进一步从个体角度研究攻击识别。试验中，8头苗猪混合饲养3 d，每天录像8 h。从视频的第一天起，就有410个攻击性的2集被标记。通过对这些片段中的每头猪进行分割，生成了820头攻击性猪和2460头非攻击性猪的子片段。然后通过水平和垂直镜像将820个侵袭性亚集扩增为2460个亚集。将4920个子集按7:3的比例划分为训练集和验证集。另外，从另外2天的视频中标记190个攻击性2秒片段，然后生成380个攻击性子片段和1140个非攻击性子片段作为测试集。首先，利用YOLOv8的识别模块对每头猪进行定位，确定其身份。其次，将时间偏移模块和深度通道注意机制融合到UniformerV2的攻击识别模块中，提高对时空特征的识别能力；最后，使用改进的DCTM-UniformerV2网络来识别猪的攻击行为。在测试集中，该算法对攻击行为的识别准确率达到97.2%。结果表明，所提出的方法可以用于识别个体猪的攻击行为。

{"title":"Aggression recognition for individual pigs based on YOLOv8 and DCTM-UniformerV2","authors":"Chen Chen , Rui Liu , Weixing Zhu , Tomas Norton","doi":"10.1016/j.biosystemseng.2025.104345","DOIUrl":"10.1016/j.biosystemseng.2025.104345","url":null,"abstract":"<div><div>The aim of this study is to develop a deep learning algorithm based on YOLOv8 and DCTM-UniformerV2 to recognise aggressive behaviour in pigs. Compared with previous studies, this study further investigates aggression recognition from individual perspectives. In the experiment, 8 nursery pigs were mixed for 3 days and then 8 h of video was recorded each day. 410 aggressive 2 s episodes were labelled from the first day of video. By segmenting each pig in these episodes, sub-episodes of 820 aggressive and 2460 non-aggressive pigs were generated. These 820 aggressive sub-episodes were then augmented into 2460 sub-episodes by using horizontal and vertical mirroring. All the 4920 sub-episodes were divided into the training and validation sets at a ratio of 7:3. Moreover, 190 aggressive 2 s episodes were labelled from the other 2 days of videos, and then 380 aggressive and 1140 non-aggressive sub-episodes were generated as the test set. Firstly, identification module of YOLOv8 was used to locate each pig and determine their identities. Secondly, temporal shift module and depth channel attention mechanism were fused into the aggression recognition module of UniformerV2 to improve the discrimination of spatial-temporal features. Finally, the improved DCTM-UniformerV2 network was used to recognise aggressive behaviour of pigs. In the test set, the aggression could be recognised by using the proposed algorithm with an accuracy of 97.2 %. The result indicates that the proposed method can be used to recognise aggressive behaviour of individual pigs.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"261 ","pages":"Article 104345"},"PeriodicalIF":5.3,"publicationDate":"2025-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145615450","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0