基於ROS2之田間載具優化設計

Abstract

隨著臺灣農業勞動力逐漸老化且人力資源日益短缺，自主田間載具已成為推動智慧農業發展之重要基礎。然而，實際農業作業環境中普遍存在地形起伏不平之特性，易導致載具行進不穩、影像震動劇烈，以及能源續航不足等問題，進而限制自主作業系統之可靠性與實用性。本研究針對上述挑戰，設計並建置一款田間自主載具，以機器人作業系統2 (Robot Operating System 2, ROS2)作為核心控制架構，並分別從行動性能、視覺穩定性與電力續航力三個面向進行系統化評估與驗證。 在行動性能方面，本研究透過滑移率(Slip Ratio)評估不同輪胎之行走效率。實驗結果顯示，在顛簸地表下，充氣胎相較於泡棉胎具有較佳之地形適應能力，能有效降低輪胎滑移現象，提升整體推進效率。在視覺穩定性方面，本研究導入二軸穩定器，以補償地形起伏所引起之車體姿態擾動。實驗結果顯示，該系統能有效抑制影像震動與視角偏移，使影像垂直位移之標準差由4.255 cm降低至2.090 cm，傾斜角標準差由4.293°降低至0.999°，提升影像品質與穩定度。在電力續航方面，所搭載之鋰電池系統在連續導控與拍攝模式下，平均可運作約157至173 min，其巡航里程足以覆蓋約25.6公畝(0.256公頃，約0.63英畝)之室外果園，能滿足中小型農田之巡檢作業需求。 綜合上述結果，本研究透過田間自主載具之系統整合與性能評估，並結合ROS2之模組化設計與通訊架構，驗證載具於行走穩定性、影像穩定性與續航能力之表現。研究結果顯示，該載具可作為精準農業監測任務之穩定作業平台，並提供農業自主載具平台設計與配置選擇之實證基礎。

With the aging of the agricultural workforce and the increasing shortage of labor in Taiwan, autonomous field vehicles have become an essential foundation for the development of smart agriculture. However, real agricultural environments are characterized by uneven terrain, which often leads to unstable vehicle motion, severe image vibration, and insufficient energy endurance, thereby limiting the reliability and practicality of autonomous field operations. To address these challenges, this study designs and implements an autonomous field vehicle using Robot Operating System 2 (ROS2) as the core control framework. The system is systematically evaluated from three aspects: mobility performance, visual stability, and power endurance. In terms of mobility performance, the slip ratio is adopted to evaluate the traveling efficiency of different tire types. Experimental results indicate that, under uneven terrain conditions, pneumatic tires exhibit better terrain adaptability than foam-filled tires by effectively reducing wheel slip and improving overall propulsion efficiency. For visual stability, a two-axis gimbal is integrated to compensate for vehicle body disturbances caused by terrain irregularities. Experimental results demonstrate that the proposed stabilization system effectively suppresses image vibration and viewpoint deviation, reducing the standard deviation of vertical displacement from 4.255 cm to 2.090 cm and the standard deviation of tilt angle from 4.293° to 0.999°, thereby improving image quality and stability. In terms of power endurance, the integrated lithium battery system achieves an average operating time of approximately 157 to 173 minutes under continuous navigation and image acquisition modes. The corresponding cruising range is sufficient to cover approximately 0.256 hectares (about 0.63 acres) of outdoor orchards, meeting the inspection requirements of small- to medium-scale farmlands. Based on the above results, this study verifies the performance of the autonomous field vehicle in terms of mobility stability, visual stability, and endurance through systematic system integration and performance evaluation, combined with the modular design and communication architecture of ROS2. The developed vehicle demonstrates its suitability as a stable operational platform for precision agricultural monitoring and provides empirical evidence to support platform design and configuration selection for agricultural autonomous field vehicles.