Elements:Open-source hardware and software evaluation system for UAV

机器人和无人驾驶飞行器（uav）的使用正在增加，它们正在成为我们日常生活的一部分。在这个过程中，机器人系统不再仅仅是一个机械设计问题，现在还包括大量的计算能力。然而，针对机器人工作负载的系统或架构研究很少，特别是在现实环境中。现有的基础设施主要集中在无人机飞行算法测试上，在配置计算能力方面缺乏灵活性。为此，该项目开发了一个软硬件协同设计框架，其中包括一个端到端的垂直集成堆栈，使可重构硬件能够被编程，并暴露在虚拟现实环境中，以解决现实的无人机导航问题。它将建立一个灵活的无人机软硬件基础设施，并为学术界和工业界的更多研究开辟令人兴奋的可能性。由于无人机现在被广泛使用，这项研究有可能在广泛的应用中产生影响。该项目开发了端到端的硬件和软件基础设施，可以评估计算需求并测试新的体系结构和技术。这包括：(1)分析无人机的工作负载，包括基于模型和基于学习的工作负载；(2)开发一个开源的软硬件无人机平台，可以用真实系统测试计算架构/系统；(3)发展可与无人机和飞行模拟对接的网络物理系统基础设施；(4)开发仿真基础设施，在运行无人机飞行场景时将高端计算卸载到其他加速器。该项目将推动机器人/无人机工作负载处理器实现的最新水平。它将为电力受限的应用平台提供新的机会，包括监视、汽车、环境、军事和灾难等。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The usage of Robots and Unmanned Aerial Vehicles (UAVs) is increasing, and they are becoming a part of our everyday lives. In the process, robotic systems have ceased to be solely a mechanical design problem but now include substantial computing power as well. However, little system or architecture research has been done for robotics workloads, especially in realistic environments. The existing infrastructure is focused on drone flight algorithm tests, with little flexibility to configure the computing capability. To this end, this project develops a software-hardware co-design framework that includes an end-to-end, vertically integrated stack that enables reconfigurable hardware to be programmed and exposed to virtual reality environments for realistic drone navigation problems. It will establish a flexible software-hardware infrastructure of drones and open up exciting possibilities for more research in academia and industry. Since drones are nowadays widely used, the research has potential for impact across a wide range of applications. The project develops an end-to-end hardware and software infrastructure that can evaluate the computing requirements and test out new architectures and technologies. This includes: (1) profiling the workload for UAVs, which include both model-based as well as learning-based workloads; (2) developing an open-source hardware/software drone platform that can be used for testing computing architecture/systems with a real system; (3) developing a cyber-physical system infrastructure that can be connected with drones and flight simulation; and (4) developing a simulation infrastructure to offload high-end computation to other accelerators while running drone flight scenarios. The project will advance the state of the art in implementation of processors for robotics/UAV workloads. It will offer new opportunities in power-constrained platforms for applications including surveillance, automotive, environment, military, and disaster to name a few.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.