【摘要】 针对现有智能行车系统在工业场景中面临的部署复杂度高、可移植性差的技术瓶颈，提出了一种基于机器人操作系统(Robot Operating System, ROS)和ORB-SLAM3算法的智能行车系统。通过功能分析，构建了智能行车系统总体架构，首先，使用深度相机实时采集料堆的环境信息；然后，对ORB-SLAM3算法进行改进，增加了稠密建图线程与安全取料线程，实现行车实时定位、料库区三维重建及目标取料点规划；最后，基于ROS的标准化运动控制接口，确保行车稳定可靠运行。模拟实验表明，智能行车系统定位与三维重建精度高，可实时生成安全高效的取料策略；ROS分布式架构与深度相机的快速部署，降低了部署复杂度，提升了智能行车系统的可移植性。研究结果有望为解决行业现存的部署复杂度高和可移植性差的问题提供切实可行的技术路径。更多还原

【Abstract】 In view of the technical bottlenecks of high deployment complexity and poor portability faced by existing intelligent crane systems in industrial scenarios, this study proposes an intelligent crane system based on the Robot Operating System(ROS) and the ORB-SLAM3 algorithm. Through functional analysis, the overall architecture of the intelligent crane system is constructed. Specifically, a depth camera is first utilized to capture real-time environmental information of material piles. Subsequently, the ORB-SLAM3 algorithm is optimized by incorporating a dense mapping thread and a safe material-picking thread, thereby enabling real-time crane positioning, three-dimensional reconstruction of the material storage area, and planning of target material-picking points. Finally, by leveraging the standardized motion control interface of ROS, the stable and reliable operation of the crane is ensured. Simulation experiments demonstrate that the proposed intelligent crane system achieves high accuracy in positioning and three-dimensional reconstruction, and is capable of generating safe and efficient material-picking strategies in real time. Furthermore, the distributed architecture of ROS and the rapid deployment of the depth camera contribute to reducing deployment complexity and enhancing the portability of the intelligent crane system. The research findings are anticipated to provide a feasible technical approach for addressing the long-standing issues of high deployment complexity and poor portability within the industry.更多还原