ABSTRACT

Autonomous transfer vehicles (ATVs) can be considered as one of the critical components of context-aware structured smart factories in Industry 4.0 era. Conventional mapping methods such as grid maps can provide information for navigation, but they are not enough for complex environments that require interactions. On the other hand, high-definition (HD) mapping, which is mainly used in traffic networks, includes more information about an environment to perform excellent autonomous behaviour. In order to increase the efficiency of ATVs in flexible factories, an up-to-date environmental map information is required to perform successful long-term autonomous navigation. Therefore, when there exists a change in the environment, a simultaneous update of HD-map is as important as the creation of it. In this study, we propose an HD-map update methodology for ATVs that operates in smart factories. To the best of our knowledge, HD mapping has not been applied in smart factories. The proposed method includes the object detection and localisation tool to detect objects visually and determines their positions in connection with the conventional maps of the environment. Experimental results of a simulated factory environment demonstrate that the ATV can properly update the HD-map when a predefined sign is removed from or a new sign is added to the environment.

摘要

自动转运车 (ATV) 可以被视为工业 4.0 时代情境感知结构化智能工厂的关键组件之一。传统的地图绘制方法如网格地图可以提供导航信息，但对于需要交互的复杂环境来说还不够。另一方面，主要用于交通网络的高清 (HD) 映射包含有关环境的更多信息，以执行出色的自主行为。为了提高柔性工厂中 ATV 的效率，需要最新的环境地图信息才能成功执行长期自主导航。因此，当环境发生变化时，高精地图的同步更新与其创建一样重要。在这项研究中，我们为在智能工厂中运行的 ATV 提出了一种高清地图更新方法。据我们所知，高清地图尚未应用于智能工厂。所提出的方法包括对象检测和定位工具，用于视觉检测对象并确定它们与环境的传统地图相关的位置。模拟工厂环境的实验结果表明，当预定义的标志从环境中移除或新标志添加到环境中时，ATV 可以正确更新高清地图。所提出的方法包括对象检测和定位工具，用于视觉检测对象并确定它们与环境的传统地图相关的位置。模拟工厂环境的实验结果表明，当预定义的标志从环境中移除或新标志添加到环境中时，ATV 可以正确更新高清地图。所提出的方法包括对象检测和定位工具，用于视觉检测对象并确定它们与环境的传统地图相关的位置。模拟工厂环境的实验结果表明，当预定义的标志从环境中移除或新标志添加到环境中时，ATV 可以正确更新高清地图。