Nano-sized unmanned aerial vehicles (UAVs) are ideal candidates for flying Internet-of-Things smart sensors to collect information in narrow spaces. This requires ultra-fast navigation under very tight memory/computation constraints. The PULP-Dronet convolutional neural network (CNN) enables autonomous navigation running aboard a nano-UAV at 19, at the cost of a large memory footprint of 320kB_ and with drone control in complex scenarios hindered by the disjoint training of collision avoidance and steering capabilities. In this work, we distill a novel family of CNNs with better capabilities than PULP-Dronet, but memory footprint reduced by up to 168× (down to 2.9kB), achieving an inference rate of up to 139frame/s; we collect a new open-source unified collision/steering 66images dataset for more robust navigation; and we perform a thorough in-field analysis of both PULP-Dronet and our tiny CNNs running on a commercially available nano-UAV. Our tiniest CNN, called Tiny-PULP-Dronet v3, navigates with a 100% success rate a challenging and never-seen-before path, composed of a narrow obstacle-populated corridor and a 180°turn, at a maximum target speed of 0.5m/s. In the same scenario, the SoA PULP-Dronet consistently fails despite having 168× more parameters.

中文翻译：

---

提炼微型超快深度神经网络用于纳米无人机自主导航


纳米级无人机（UAV）是飞行物联网智能传感器在狭窄空间收集信息的理想选择。这需要在非常严格的内存/计算限制下进行超快速导航。 PULP-Dronet 卷积神经网络 (CNN) 能够在 19 纳米无人机上运行自主导航，但代价是 320kB 的大内存占用，并且复杂场景中的无人机控制因防撞和转向能力的不相交训练而受到阻碍。在这项工作中，我们提炼出了一个新颖的 CNN 系列，其功能比 PULP-Dronet 更好，但内存占用减少了高达 168 倍（低至 2.9kB），推理速率高达 139 帧/秒；我们收集了一个新的开源统一碰撞/转向 66images 数据集，以实现更强大的导航；我们对 PULP-Dronet 和在商用纳米无人机上运行的微型 CNN 进行了彻底的现场分析。我们最小的 CNN，称为 Tiny-PULP-Dronet v3，以 100% 的成功率导航一条具有挑战性且前所未见的路径，该路径由狭窄的障碍物密集的走廊和 180° 转弯组成，最大目标速度为 0.5多发性硬化症。在相同的场景中，尽管 SoA PULP-Dronet 的参数增加了 168 倍，但它始终失败。