精微運動系統實驗室
國立中興大學 機械工程所
實驗室由陳昭亮副教授帶領，以機電系統整合開發、無人載具、機械學習的實際應用為研究路線。實驗室有多種微電腦、單晶片控制器、感測器、驅動器、制動器…等電子設備供學生進行相關實作練習。除此之外，還有各式電子零件、專業工具組、容易拆裝的結構件，讓學生能夠方便、自由地創造出心中理想的機器人。
在此除了訓練學生機電設備整合能力以外、也能練習將課堂所學之概念，應用於控制系統之設計，在實作過程中培養學生發現問題、解決問題的能力。實驗室行事風格偏向設定目標、自由發揮，學生除了實驗室研究主線外，也可自由選擇其他研究路線。
本實驗室近年以「無軌式自走車」、以及「強化式學習在足型機器人上的應用」等為主研究項目。「無軌式自走車」為產學合作計畫，將合作廠商所開發的感測器應用在自走車上，並研究如何整合其他感測器以彌補其不足，使得自走車可以在更多樣的環境下正常運作。此研究主要屬於感測器整合應用、自動控制系統開發、多種演算法整合應用的部分。
「強化式學習在足型機器人的應用」方面為實驗室近幾年開始進行的項目，此項目牽涉到機械學習在實際機電系統的應用，如何將模擬結果轉移到實體機台並使其能夠正常運作是本項目最大挑戰。研究此項目所涵蓋的範圍也最為廣泛，較具挑戰性。在此能夠培養機電系統整合、自控系統開發、機動學/動力學/自動控制…等相關學術應用、機械學習學理/實作能力。
實驗室介紹：
Figure 1
實驗室自行搭建的無軌式自走車
The Guideless automatic vehicle built in the lab.
Figure 2
用於強化式學習的機械足Robot foot used in “Application of Reinforcement Learning on Foot Robot”
Lab ID: 931
Host: Associate Professor, Jau-Liang Chen
Main research topic:
Guideless Automatic Vehicle
Application of Reinforcement Learning on Foot Robot
Brief introduction:
The lab.931 which is led by professor Jau-Liang Chen, features in development of electro-mechanical system, unmanned vehicle, and the application of machine learning. There are many microcomputers, single-chip controllers, sensors and actuators for students to practice. Moreover, professional tools, various electric components and easy disassembly parts, make it convenient to build ideal robot in the lab.
Here, students can not only learn the skills of combination of mechanics and electrics, but also the application of the concept we learned on the class. Brings up the ability that find and fix the problem during implementation.
In recent years, there are two main topics in the lab. “Guideless Automatic Vehicle” and “Application of Reinforcement Learning on Foot Robot“. The “Guideless Automatic Guided Vehicle” project is an industry-academia collaborative project. In this project, we install the product of the CO-OP company on our Guideless Automatic Guided Vehicle, it is required to build up the whole control system of the Guideless Automatic Guided Vehicle. More, the disadvantage of the product has to be cover by other solutions. So that our Guideless Automatic Guided Vehicle work normally in various environments. The research matters of this project contain sensor fusion, development of automatic system, and usage of multiple algorithms.
“Application of Reinforcement Learning on Foot Robot“ is a relatively new project, that was started in the last two years. It refers to the methods of applying machining learning on real robot. The biggest challenge of the project is the adaption of the result in simulative world into the real one. Various research areas are included in this project, makes it a lot challengeable compare with the other.