包傑奇Jacky Baltes李玟諺Lee, Wen-Yen2024-12-179999-12-312024https://etds.lib.ntnu.edu.tw/thesis/detail/2639cc07ca55ab26488ec447381144cd/http://rportal.lib.ntnu.edu.tw/handle/20.500.12235/122911在台灣，機車是人們通勤的主要工具之一。與此同時，隨著人工智慧的快速進步，仿人機器人已經成為未來的趨勢。為了促進仿人機器人的發展，我們進行了研究以探索其可行性。在這項研究中，我們的目標是控制機器人騎機車並通過台灣的駕照考試。為了完成這項任務，我們需要解決騎機車的最基本問題——平衡。在我們的研究中，我們實施了模型預測控制（MPC）來進行自平衡測試。同時，我們將討論兩輪車的建模、MPC優化算法和機器人運動規劃的逆運動學。為了評估可行性，我們還使用了PID控制器進行比較。最後，我們展示了結果，證明選擇MPC作為我們主要方法的優勢。In Taiwan, scooters are one of the primary tools for people to commute. Simultaneously, with the rapid advancements in artificial intelligence, humanoid robots have already become a trend for the future. To promote the development of humanoid robots, we have undertaken research to explore their feasibility. In this study, we aimed to control a robot to ride a scooter and pass the Taiwanese driving license test. To achieve this task, we needed to solve the most fundamental issue of riding a scooter—balance. In our research, we implemented Model Predictive Control (MPC) to conduct the self-balancing test. Simultaneously, we would discuss about modeling for the two-wheeled vehicle, MPC optimization algorithm and robot motion planning with inverse kinematic. To evaluate feasibility, we also used a PID controller for comparison. Finally, we present the results, demonstrating the advantages of choosing MPC as our primary method.noneHumanoid RobotsTwo-wheeled VehiclesClassical ControlRobot Motion PlanningNeural Network學術論文