Agricultural robotic agents have been becoming useful helpers in a wide range of agricultural tasks. However, they still heavily rely on manual operations or fixed railways for movement. To address this limitation, the AgriVLN method and the A2A benchmark pioneeringly extend Vision-and-Language Navigation (VLN) to the agricultural domain, enabling agents to navigate to the target positions following the natural language instructions. We observe that AgriVLN can effectively understands the simple instructions, but often misunderstands the complex ones. To bridge this gap, we propose the T-araVLN method, in which we build the instruction translator module to translate noisy and mistaken instructions into refined and precise representations. When evaluated on A2A, our T-araVLN successfully improves Success Rate (SR) from 0.47 to 0.63 and reduces Navigation Error (NE) from 2.91 m to 2.28 m, demonstrating the state-of-the-art performance in the agricultural VLN domain. Code: https://github.com/AlexTraveling/T-araVLN.
The scale and intensification of poultry production continue to increase, and Internet of Things (IoT) technology has been widely adopted in poultry farming. Along with the increasing IoT adoption comes the need to deal with the issues of real-time processing, analysis and application of massive field data. Time series data in large-scale egg production have the characteristics of multiple collection equipment or sensors, large datasets, and high-speed throughput. This research describes a monitoring and alarm system based on time series database and alarm model. The system uses distributed stream computing technology to process various time series data at high speed. The time series database “OpenTSDB” is used in the system to storage and query the time series data efficiently. The write performance of OpenTSDB is scalable, which can reach a write performance of one million qps, and the query delay is less than 100ms in 99% of the cases. Multiple alarm strategies were developed, allowing the system to alarm as single-indicator or indicator segment and enabling application in more scenarios. The monitoring and alarm system is expected to significantly advance the ability of modern poultry producers in their efficient monitoring and processing of big data and real-time alarm of the flock performance and housing conditions. Keywords: Time Series Data; OpenTSDB; Distributed Stream Computing; Laying-hens
The issues of inadequate digital proficiency among agricultural practitioners and the suboptimal image quality captured using mobile smart devices have been addressed by providing appropriate guidance to photographers to properly position their mobile devices during image capture. An application for crop guidance photography was developed, which involved classifying and identifying crops from various orientations and providing guidance prompts. Three steps were executed, including increasing sample randomness, model pruning, and knowledge distillation, to improve the MobileNet model for constructing a smartphone-based orientation detection model with high accuracy and low computational requirements. Subsequently, the application was realized by utilizing the classification results for guidance prompts. The test demonstrated that this method effectively and seamlessly guided agricultural practitioners in capturing high-quality crop images, providing effective photographic guidance for farmers.