Improving Autonomous Nano-Drones Performance via Automated End-to-End Optimization and Deployment of DNNs

Abstract

The terms and conditions for the reuse of this version of the manuscript are specified in the publishing policy. For all terms of use and more information see the publisher's website.This item was downloaded from IRIS Università di Bologna (https://cris.unibo.it) When citing, please refer to the published version.

Towards a Multi-Pixel Time-of-Flight Indoor Navigation System for Nano-Drone Applications

“…Moreover, carrying heavy and power-hungry 3D scanners, such as stereo cameras and LiDARs, is not considered a limitation for conventional robotic applications [35]- [37]. However, these assumptions do not hold for miniaturized and low-power robotic platforms, where the hardware cost is a concern, the payload is limited to a few tens of grams, and the computation power budget is limited to hundreds of mW [7], [11], [38]. Hence, enabling onboard mapping on this tiny class of devices is still an open problem.…”

“…Nano-robots, and especially palm-size UAVs, weigh only a few tens of grams and benefit from increased agility compared to their standard-size counterparts, enabling them to fly in narrow spaces reliably [5], [6]. Furthermore, their reduced dimensions make nano-UAVs perfect candidates for safely operating near humans, especially in cramped indoor environments [7], [8]. In most practical applications, the mission of the nano-UAV is to follow a path through the environment that is predefined or V. Niculescu and L. Benini are with the Integrated Systems Laboratory of ETH Zürich, ETZ, Gloriastrasse 35, 8092 Zürich, Switzerland (e-mail: vladn@iis.ee.ethz.ch, lbenini@iis.ee.ethz.ch).…”

“…Benini is also with the Department of Electrical, Electronic and Information Engineering of University of Bologna, Viale del Risorgimento 2, 40136 Bologna, Italy. adjusted dynamically during the mission [5], [7]. For instance, finding the source of gas leaks [9] or localizing and reaching sensor nodes for data acquisition [10] are only a few examples of such applications.…”

“…With the aid of such sensors, recent works demonstrated SLAM on nano-UAVs, but only under the assumption that the complex SLAM could be offloaded to an external base station [19]. This approach reduces the flight time due to the significant power consumption introduced by the radio communication with the base station [7]. Even more serious issues are the latency associated with the wireless communication protocol and the limited radio link range, which typically constrains the operating area within a few tens of meters in indoor environments [7].…”

“…This approach reduces the flight time due to the significant power consumption introduced by the radio communication with the base station [7]. Even more serious issues are the latency associated with the wireless communication protocol and the limited radio link range, which typically constrains the operating area within a few tens of meters in indoor environments [7]. Furthermore, because of the limited measurement capabilities of the early ToF sensors (i.e., a single distance value per sensor and a narrow FoV) [5], the existing systems that enable SLAM with nano-UAVs can only map simple-geometry environments such as long flat corridors.…”