Impatto dell'ottimizzazione delle risorse sul QoS della Crazyflie AI Deck 2.1

Miniaturized autonomous drones with artificial intelligence capabilities are increasingly in demand and widespread, their size, reduced energy and computing availability make it difficult to integrate AI-based navigation algorithms, which require higher computing capabilities. Bitcraze proposed this challenge, Crazyflie 2.1 a nano drone of only 27g, which has the possibility of hosting AiDeck, an additional board that allows the execution of AI algorithms while consuming a few mW. In this thesis work, we will discuss the computing capabilities of the AiDeck, focusing on the computing performance. In the next chapters, we will therefore analyze the crazyflie 2.1 system, its development system, the functioning of AIDeck, and how the performance of gap8, the AiDeck processor, thanks to its parallel computing capabilities and its hardware accelerations, impacts on the Crazyflie QoS. The work is developed in four chapters: • Chapter 1 describes the Crazyflie 2.1 and some of its extensions. • Chapter 2 exploring gap8, the AiDeck main processor. • Chapter 3 introduce the software development kit for gap8, the overview, the installation, the tool. • Chapter 4 Describe the test used to benchmarking processsor, the results, the impact of gap8 hardware accelleration. The following installations and settings, and hardware components have been used in this thesis work: • Bitcraze Crazyflie 2.1 drone basic version, see chapter 1 • Bitcraze AiDeck 1.1, see chapter 1.2.4 • Olimex arm-usb tiny jtag interface [26] • Canonical Ubuntu 22.04 desktop operating system [25] • GreenWaves software development kit repository [20] • Bitcraze crazyflie firmware repository [12] • Anaconda Software Distribution platform to develope in python [8] • Python programming language version 3.8 [24]