Please use this identifier to cite or link to this item:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/73092
Title: | 智慧割草機設計與建製 Design and Implement of an Intelligent Robotic Lawn Mower |
Authors: | Kai-Sheng Wang 王凱陞 |
Advisor: | 黃振康(Chen-Kang Huang) |
Keyword: | 阿卡曼轉向系統,路徑規劃,感測融合,即時定位與地圖建置,自駕車建置,Robot operating system (ROS), Ackermann steering system,Path planning,Sensor fusion,Simultaneous localization and mapping (SLAM),Self-driving vehicle establishment,Robot operating system (ROS), |
Publication Year : | 2019 |
Degree: | 碩士 |
Abstract: | 本研究由遙控車體進行改裝,探討遙控車體的轉向系統-阿卡曼轉向系統,並以此轉向系統之特性計算車體的運動狀態,進行車體運動軌跡的追蹤,測試不同的PWM訊號所對應的車體速度、角速度關係式。以Robot Operating System (ROS)進行整體系統的實作和維護,並以霍爾感測器測試輪胎轉速進行直線速度的感測,實驗得到霍爾感測器所感測的速度與實際測量的平均速度偏差值為0.7 m/s。透過Kalman Filter進行感測融合,達到更準確的車體狀態追蹤,由於轉向上僅使用IMU進行感測,誤差會隨著時間逐漸增加,因此另外透過環境資訊進行車體位置的定位。透過Intel D435深度相機的深度資訊,作為智慧割草機對環境資訊的主要感測工具,搭配Kalman Filter所估算的車體狀態,以Particle Filter演算法,進行車體位置的定位,確認了車體狀態追蹤和定位後,便可搭配環境資訊對整體環境進行建圖,即可達到同步定位與建圖Simultaneous Localization And Mapping(SLAM)的實作。
本研究將路徑規劃分為全域性和區域性的路徑規劃,運用最短路徑演算法(A* search)和區域動態演算法(Dynamic window approach)的融合,達到基本的指定位置移動的效果,且具備遠端遙控並模擬監測的功能。並以模擬的方式進行區域掃蕩的演算法實現,透過模擬的方式測試如何將一個不規則形狀的區塊,完整的掃蕩過一輪,並以視覺化的介面進行實作。 除上述軟體實現之外,本研究也包含了如何將遙控車逐漸建置成具備割草機構的智慧平台自駕車,並設計割草機構進行割草測試,以電流感測器測試割草情形與驅動馬達的電流回饋追蹤,藉此達到割草情形的監控。 This study is dedicated to establish an intelligent lawn mower. Based on a remote control car body and calculate the motion state of the mower body based on the characteristics of Ackermann steering system. The corresponding mower body speed and angular velocity relationship are implemented and maintained by the Robot Operating System (ROS). The Hall sensor is used to test the tire’s rotation speed to perform linear speed sensing. The measured speed from Hall sensor and the actual measured average speed deviation are 0.7 m/s, and sensor fusion is implemented by Kalman Filter, so as to achieve more accurate mower state tracking, since the steering is only using the IMU for sensing, error gradually increase overtime, so it is necessary to additionally locate the position of the mower through environmental information. Through the depth information of the Intel D435 depth camera, as the main sensor of the intelligent mower's environmental information, with the mower state estimated by Kalman Filter, using the Particle Filter algorithm to locate the mower position, confirm the mower After state tracking and positioning, the overall environment can be mapped with environmental information to achieve Simultaneous Localization And Mapping (SLAM). In this study, the path planning is divided into global and local path planning, implement by the combination of the shortest path algorithm (A* search) and DWA (Dynamic window approach). Achieve the basic positional movement effect、remote control and analog monitoring functions. In the field of region filling, tests how to completely fill an irregular shape map through simulation. In addition to the above software implementation, this study also includes how to gradually build a remote control car into a self-driving vehicle with a mowing mechanism, and design a mowing mechanism’s prototype, and test the mowing with a current sensor. The current feedback tracking of the drive motor is used to monitor the mowing situation. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/73092 |
DOI: | 10.6342/NTU201901401 |
Fulltext Rights: | 有償授權 |
Appears in Collections: | 生物機電工程學系 |
Files in This Item:
File | Size | Format | |
---|---|---|---|
ntu-108-1.pdf Restricted Access | 9.59 MB | Adobe PDF |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.