基於UWB定位的啟發式無人機編隊控制設計

Abstract

本論文提出了一套由三架無人機、Ultra-wide band (UWB)、GPS及Raspberry Pi 3B+ (RPi3) 所組成的無人機群協同飛行系統架構。 研究過程中嘗試以UWB來支援無人機編隊的相對定位，但因定位效果不如預期，所以仍採用GPS進行定位。此架構可自動控制僚機維持特定任務隊形，跟隨長機協同飛行。樹梅派透過Wi-Fi建立多架無人機之間的區域無線網路，並用UDP通訊協定讀取來自無人機飛控電腦中的空速計、氣壓計、陀螺儀等各項儀器的飛航資訊，後經由提出的PI自動控制器，可以將兩架跟隨無人機在飛行期間的水平、垂直誤差都控制在10 cm以內，再經由無人機之間的通訊管道，組織成Leader-Follower結構的無人機編隊控制系統，使無人機群跟隨誤差達到公分級的精準度。同時本論文提出一套模擬器，將無人機的各項相關數據以及相關環境參數納入考量，搭配前述所提出之PID控制方法，以三維立體動畫的方式預測無人機群的飛行，可從更多角度觀察無人機的模擬狀況，盡可能的避免實際飛行時的危險情況。

The paper proposes a collaborative flight system architecture composed of three drones, Ultra-wide band (UWB), GPS, and Raspberry Pi 3B+ (RPi3). This architecture automatically controls the followers to maintain mission formations and fly with the leader. During the research process, UWB was tried to support the relative positioning of UAV formations, but because the positioning effect was not as good as expected, GPS was still used for positioning. Due to the high mobility of UAV, we can adjust the formation of the UAV swarm in GPS-denied environments using UWB positioning signals. During flight, the Raspberry Pi establishes a communication channel through a Wi-Fi local wireless network formed among UAV swarm. UDP communication protocol is used to allow UAVs to share their information with each other to determine and adjust their flight attitudes. The Raspberry Pi also reads flight information from the flight control computers of UAV. Through the proposed PI automatic control method, the horizontal and vertical errors of the two follower are controlled within 10 cm. Then, a Leader-Follower structured UAV formation control system is organized, achieving centimeter-level precision in following errors for the drone group. Additionally, the paper proposes a simulator that takes into account various relevant data of UAV and environmental parameters before the actual flight of the UAV swarm. With the control method proposed above, the simulator predicts the flight of the UAV swarm and presents it with three-dimensional animations,allowing observation of UAV simulations from various perspectives, minimizing potential risks during actual flights.