The evolution of energy-efficient ultra-low-power (ULP) parallel processors and the diffusion of convolutional neural networks (CNNs) are fueling the advent of autonomous driving nano-sized unmanned aerial vehicles (UAVs). These sub-10cm robotic platforms are envisioned as next-generation ubiquitous smart-sensors and unobtrusive robotic-helpers. However, the limited computational/memory resources available aboard nano-UAVs introduce the challenge of minimizing and optimizing vision-based CNNs – which to date require error-prone, labor-intensive iterative development flows. This work explores methodologies and software tools to streamline and automate all the deployment of vision-based CNN navigation on a ULP multicore system-on-chip acting as a mission computer on a Crazyflie 2.1 nano-UAV. We focus on the deployment of PULP-Dronet (Palossi et al., 2019), a state-of-the-art CNN for autonomous navigation of nano-UAVs, from the initial training to the final closed-loop evaluation. Compared to the original hand-crafted CNN, our results show a 2×2\times reduction of memory footprint and a speedup of 1.6×1.6\times in inference time while guaranteeing the same prediction accuracy and significantly improving the behavior in the field, achieving: i) obstacle avoidance with a peak braking-speed of 1.65m/s and improving the speed/braking-space ratio of the baseline, ii) free flight in a familiar environment up to 1.96m/s (0.5m/s for the baseline), and iii) lane following on a path featuring a 90deg turn – all while using for computation less than 1.6% of the drone’s power budget. To foster new applications and future research, we open-source all the software design in a ready-to-run project compatible with the Crazyflie 2.1.

中文翻译：

---

通过自动端到端优化和 DNN 部署来提高自主纳米无人机性能


高能效超低功耗 (ULP) 并行处理器的发展和卷积神经网络 (CNN) 的普及正在推动自动驾驶纳米级无人机 (UAV) 的出现。这些小于 10 厘米的机器人平台被设想为下一代无处不在的智能传感器和不引人注目的机器人助手。然而，纳米无人机上可用的计算/内存资源有限，带来了最小化和优化基于视觉的 CNN 的挑战——迄今为止，这需要容易出错、劳动密集型的迭代开发流程。这项工作探索了方法和软件工具，以简化和自动化 ULP 多核片上系统上基于视觉的 CNN 导航的所有部署，充当 Crazyflie 2.1 纳米无人机上的任务计算机。我们专注于 PULP-Dronet（Palossi 等人，2019）的部署，这是一种用于纳米无人机自主导航的最先进的 CNN，从最初的训练到最终的闭环评估。与原始手工制作的 CNN 相比，我们的结果显示内存占用减少了 2×2 倍，推理时间加速了 1.6×1.6 倍，同时保证了相同的预测精度并显着改善了现场行为，实现了：i）避障峰值制动速度为1.65m/s，提高基线的速度/制动空间比，ii）在熟悉的环境中自由飞行高达1.96m/s（0.5m/s为基线），以及 iii） 90 度转弯路径上的车道跟踪 - 所有这些都用于计算，小于无人机功率预算的 1.6%。为了促进新的应用和未来的研究，我们在与 Crazyflie 2.1 兼容的现成运行项目中开源了所有软件设计。