瑞典初创公司Scaleout Systems利用Nvidia Jetson Orin Nano处理器开发了一种能够自主识别和攻击目标的无人机系统，该系统在NATO的ALMA项目演示中成功实现了无需人类输入、无外部通信的自主作战能力1。该系统采用YOLOv8 Nano目标检测模型进行本地边缘计算，整个任务在320秒内完成，其中侦察阶段约200秒1。
在瑞典冬季的测试中，该系统在-18℃的雪地环境下运行，实现了30帧每秒的帧率和30毫秒的目标延迟，展现出强大的自主能力和抗干扰性能1。Scaleout Systems利用Scaleout Edge和联邦学习技术构建战术计算机视觉网络(TCVN)1。其FEDAIR项目获得NATO DIANA加速器提供的10万欧元资金支持1。
联合国专家组最近发布报告指出，符合战争法律的自主武器需要人类判断和控制，相关建议将在11月于日内瓦进行审议1。
Swedish startup Scaleout Systems has developed an autonomous target-selection drone system powered by Nvidia's Jetson Orin Nano chip, demonstrating the capability to identify and attack targets without human intervention or external communication during NATO's ALMA project trials.1 The system employs a compact AI model running locally on the edge device for onboard processing, utilizing the YOLOv8 Nano object detection model on the Jetson Orin Nano processor.1
Testing conducted in Swedish winter conditions at temperatures as low as −18 degrees Celsius (0 degrees Fahrenheit) achieved a frame rate of 30 frames per second with a target detection latency of 30 milliseconds, with the entire mission completed in 320 seconds, including approximately 200 seconds for reconnaissance.1
The platform leverages Scaleout Edge technology and federated learning to construct a tactical computer vision network (TCVN).1 Scaleout Systems' FEDAIR project has secured 100,000 euros in funding support from the NATO DIANA accelerator.1 A recent report by UN experts emphasized that compliance with the laws of war requires human judgment and control, with recommendations scheduled for review in Geneva in November.1
评论
还没有评论，欢迎留下第一条。