Abstract:Livestock farming operation robots are important technological carriers for smart livestock farming and play significant roles in alleviating labor shortages, improving production efficiency, enhancing biosecurity, and improving animal welfare. With the continued advancement of large-scale and intelligent livestock production, related equipment and research have gradually evolved from single-task mechanized assistance to perception, task execution, and collaborative applications in complex farming scenarios. Owing to the diversity of task objects, dynamic changes in animal states, and significant disturbances in barn environments, livestock farming operation robots still face substantial challenges in perception, navigation, execution, decision-making, and coordinated control. It systematically summarized recent advances in key technologies, including multimodal perception and scene understanding, mobile platforms and navigation, compliant execution, and device-edge-cloud collaboration. It also reviewed their application development in typical scenarios such as precision feeding, product harvesting and collection, health and behavior monitoring, reproductive assistance, and barn cleaning, and further identified the main bottlenecks, including insufficient perception robustness, difficulties in live-animal modeling and compliant control, limited real-world generalization, insufficient multi-device collaboration, inadequate consideration of animal welfare constraints, and limited engineering applicability and scalability. Overall, the key to future development lies in overcoming technological bottlenecks in robust perception under high interference, autonomous decision-making in dynamic environments, and low-stress interactions among humans, machines, and animals.