摘要
记者从发明人提供的背景资料中获得以下引述：“稳定和可靠的机器人系统正变得越来越普遍，这有助于促进联合国载人系统技术的最新进展和扩散。这些技术可以地基系统、基于ACRI的系统和/或基于海洋的系统。在许多情况下系统配备有一个或多个摄像机、一个或多个麦克风和/或其他记录设备。来自无人驾驶系统（s）的图像/声音流可由操作员控制使操作员知道地面上发生了什么的装置。在许多情况下它可能是有用的使影像流不与影像/音讯流内侦测到的不同物件相对应。目前，图像/声音帧内的o个物体通常使用机器学习模型来检测。这些机器学习模型通常托管在具有专用数据源的专用硬件上。然而，在操作场景中，使用第三方模型（例如，来自那些能够培训这些模型，因为这些实体拥有所需的培训数据。此外，操作RS字段中可能希望选择需要为特定操作检测哪些对象类型，因为指示其他对象类型可能会干扰操作员评估操作情况的能力。因此，在无人驾驶飞行器或具有有限计算r的另一设备上启用这种功能资源充其量也是具有挑战性的。
Abstract
Reporters obtained the following quote from the background information supplied by the inventors:“Stable and reliable robotic systems are becoming increasingly common, which has contributed to therecent advancement and proliferation of un manned system technologies. These technologies can beground-based systems, acri al-based systems, and/or maritime-based systems. In many instances thesesystems are equipped with one or more cameras, one or more microphones, and/or other re cordingequipment. The image/sound stream from the unmanned system(s) may be sen t to an operator’s controldevice to make the operator aware of what is happenin g on the ground. In many cases it may be usefulfor the image stream to be annot ated with different objects detected within the video/audio stream.Currently, o bjects within images/sound frames are generally detected using machine learning models.These machine learning models are generally hosted on dedicated hardware with dedicated data sources.However, in operational scenarios it may be critic al to use third-party models (e.g., from entities that areable to train those m odels because those entities have the required training data). Moreover, operato rsin the field may want to choose which object types need to be detected for a particular operation becauseindicating other object types may interfere with th e operator being able to assess the operational situation.Thus, enabling this f unctionality on an unmanned vehicle or another device with limited computation resources is challenging at best.”
NETL
NSTL