抽籤機
此篇報告中,主要是在敘述以單晶片8051為主體,運用其所具有之功能,製作成此作品,稱為記數抽籤機,又稱抽籤機,根據本篇報告結果,可歸納出以下結果:1.本實驗能讓初學者對單晶片8051之特性與功能有更深的了解,以增進實驗與成品實作的能力。2.此作品以便利為原則,故其體積不大,重量輕攜帶方便。3.使用之元件普遍,價格便宜,在電子材料行均可買的到,成本低。4.依本品之功能,可用於學校讓老師進行抽問,也可用於其他有抽籤或抽獎活動之場合,公平又公正。;In the report, we take the single chip 8051 as a main body, making use of function to do the counting drawing lots machine, which are also called drawing lots machine. According to the resultof this report, conclusions are as followings: 1. With the experiment, the beginners egt deeper understanding about the characteristics and the finished products. 2. This machine is light and its volume is small, so it is easy and convenient to carry with. 3. The essential elements are common,cheap and can easily be bought in the store which is sellong electronic materials. 4. The drawing lots machine can be used at schools when teachers want to pick out some students to ask or it can be used for some drawing lots occasion as well. The outcome is just and fair.
打破語言隔閡-以圖點餐之數位餐廳
本次研究利用Web Sevrice軟體系統和Microsoft SQL Server的全文搜尋(Full-Text search)並藉由使用者選擇的圖片組合排列從系統資料庫中檢索比對，篩檢出正確或者最相似的餐點。Web Service 是建立於網際網路上之系統架構，且為一個具開放性、分散式的軟體元件，其基礎建立在HTTP、XML、SOAP、WSDL和UDDI等標準協定上，此研究即是透過Web Service系統與資料庫伺服器的連結進行。很多數位化資訊是以沒有結構化的文字資料，存成純文字檔及格式化的文件。這一類的文字資料儲存在檔案系統，而許多組織現在將這一類資料儲存到關聯式資料庫。這樣做就可以廣泛利用資料庫的好處-包括高效能存取、查詢能力、對使用者而言簡單的應用程式使用者介面，以及安全的遠端存取、查詢能力、對使用者而言簡單的應用程式使用者介面，以及安全的遠端存取。運用此功能即可在配對完成時搜尋最佳的結果並加以排名。「打破語言隔閡-以圖點餐之數位餐廳 」不只消除語言隔閡，更讓點餐手續方便許多，並讓錯誤的可能性降低。另外圖片資料庫的增減簡單快速、介面明瞭易懂。希望這些應用及創新能讓人們用餐方便及餐廳事業蓬勃發展。This research uses Web Service software system and Microsoft SQL Server. To choose the correct or the most similar meals by checking the combination and arrangement of the pictures chosen by users form the system data banks. Web-Service is a system structure establishes at the Internet. And its an open, dispersed software part. Its foundation is established at the standard agreement of HTTP,XML,SOAP,WSDL, and UDDI. The research proceeds by linking Web Service system and data-bank server lots of digital data are stored as words files and stylization documents without any structural words data. This kid of ward data are stored at file system, and many organized systems now store them at connected data banks. By doing this, people can make the most advantage of data banks- including efficient access, inquiry abilities, simple applies user’ interface, and safe remote access. People can find the best results and place them when this function finishes matching. It can not only eliminate language barriers but also make order procedure more convenient. It can also lower the mistake possibilities. In addition, we can increase or decrease pictures of data banks easily and fast. We can make the interface understood easily. We hope the applications and innovations can make it more convenient for people to order meals. And the business of restaurant can be developed successfully.
影像的色偏校正
The human eyes can adjust automatically the color of object under different illuminants. But digital camera does not have such mechanism, must join the white balance to simulate the color constancy of human eyes. There are many existing white balance methods. They can be categorized into two types. The first type utilizes the widespread assumption on the dealing natural scenery, which has the advantage of simplicity. Another type explores knowledge of the semantic content, which has the advantage of accuracy. In this study we modify and enhance the widespread assumption methods that can adopt the advantage of the simplicity and accuracy. Our proposed method is structured in two main parts: a color cast detector and a color cast remover. The detector first analyzes the color distribution of the image with simple statistical tools to determine whether or not the image has color cast. The remover, a modified and enhanced version of the widespread assumption methods ( gray world and max RGB ), is then applied on the color cast image. From the experiment results, it demonstrates the efficacy and performance of the proposed method.人類的視覺能夠自動修正因光線變化而改變的物體顏色，但是數位取像設備的感光元件卻不具有這樣的功能，必須加入白平衡的功能，才可模擬人眼維持色彩恆常的特性。現有的白平衡演算法可分為兩大類型，第一類型為廣泛假設型，具有運算速度快、與取像設備無關的特性，但是平均誤差會較大；另一類型為預知特性型，其特徵為準確性較高，但是運算較速度較慢、建立色彩特性資訊時所需的成本較高，本篇研究針對廣泛假設型的演算法做些修正與增強，使其具有較小的平均誤差以獲得較佳色彩品質的影像。我們將影像的色偏校正分為色偏偵測與色偏移除等兩個階段來進行，由於僅有被偵測出有色偏的影像才需要進行色偏的移除，所以可以避開將無色偏或固有色偏的影像做錯誤的修正，由實驗的結果看來，我們的方法確實有效，除了具有運算速度快、與取像設備無關的優點外，其色偏校正能力也較現有的方法好，使得影像色彩更能與人眼所見相近。
記憶學習機
一般學生對於學校課程學習的負擔，再加上對本身的自信心不足，往往導致學生自己所「背」的東西不是十分確信是否有真正記進腦中，也常常困擾著他們。因此，引發本小組想設計一個可以自我評測且可以立即得到成績的輔助學習機;在研究過程中，本小組設計記憶學習機不但可以做到使用者自我評測的功能，還可以讓使用者馬上得到檢測的結果並且使用者可以回顧先前成績欄紀錄;記憶學習機經過本小組組員的辛苦孕育下誕生，本小組便拿著「它」讓班上同學認識，雖然「它」不是同學心中的最佳主角，但是同學卻一致認定「它」是『最佳伴讀夥伴』。本作品主要功能有:1．能調整使用者所要求的測試時問。2．該作品能記憶此次測試成績，以便和下次相互比較。3．依使用者的需求選擇中文、英文、數字及遊戲模式做記憶訓練。4．能有立即性的成效及回饋。 Due to the pressure of courses and lack of confidence, many students are often not sure of what they have recited on a lesson previously, which in turn always disturbs and cut down learning confidence of students. To eradicate the obstacles they encounter, we decide to research into this topic. A memory-based learning-aided machine is designed for students to self-evaluate themselves and to get the feedback at once. In the process of development, the memory-based learning-aided machine not only lets users practice the exercise of their own, but also is able to derive the outcomes immediately and retrieve the previous records effectively and efficiently. We strive to make this learning-aided machine at its best performance as possible as we can. Besides, the memory-based learning-aided machine provides the alternatives for the users to answer the question in a way that they like. For example, users can answer questions in English mode, can choose the limit time into test themselves in a time. In addition, the system can expand its material by inputting data into the knowledge base. By our group members' lasting efforts, the learning-aided machine was created, finally. According to the result of experiment that we applied to the classmates, we conclude that the memory-based learning-aided machine is the best assistant and learning partner for the students. We would like to introduce it to classmates and hope they will be interested in using it to enhance their learning motivation and performance.
MTU值與網路效率的關係
中文摘要：\r 本研究旨在探討，在TCP/IP 協定之下，最大網路傳輸效率之MTU\r （Maximum Transmission Unit,最大傳輸單位）值。根據IEEE 所公\r 佈Ethernet 之標準，經實地模擬實驗，本研究之結論如下：（1）在\r 單純兩台電腦直接串接時，由於幾乎無任何干擾與封包的碰撞，可設\r 定之MTU 最大值以1500bytes 為最佳；（2）但在模擬真實的大型廣域\r 網路時，由於碰撞增加及雜訊增多，經模擬實驗顯示，MTU 值為\r 1500bytes 時無法在資料傳送、接收及重組時取得平衡，最佳之MTU\r 值落於500 到600bytes 之間，進一步研究顯示MTU 值設在500 到\r 600bytes 時可提升傳輸效率約50%。同時，本研究亦討論將此機制應\r 用於未來新網路標準之可行性及其必要性。Abstract：\r The main purpose of this research is to explore the best network\r transmission efficiency's Maximum Transmission Unit under TCP/IP\r protocol. According to the Ethernet's standard of IEEE and simulated\r experiments, the outcome of this research are :( 1) When connecting two\r computers serially, there is almost no interference and impact of packets,\r the best MTU that we can set up is 1500bytes ;( 2)When simulating the\r truly wide area network, interference and impact are rising. The result of\r this simulates experiment shows that when MTU is at 1500bytes, it's\r unable to keep balance when sending, receiving and re-composing. The\r best new MTU is between 500bytes and 600bytes. And when MTU is\r between 500bytes and 600bytes, the network transmission efficiency can\r be promoted about 50% higher. At the same time, the research is also\r discussing the feasibility and necessity of applying it to the standard of\r the future network.
Rubik's Cube Solver
Aim: Over the years I became quite quick at solving the cube. I was keen to see if I could create a mechanical system that would do it in a similar time. Because of financial limitations and equipment I thought it impossible to achieve my usual times of around 1 minute and so settled on a target of 10 minutes. So my aim became; “To create a mechanical system that could solve the cube 100% reliably in less than 10 minutes” What I did: I started from the view that I wanted to get it to find a solution using the process that I usually use. The downside of this approach was that this approach meant that most internet research was irrelevant to my project. Also some methods I found were very sophisticated and expensive eg. the university professor who created a system to solve it in 6 seconds. I wrote software capable of solving the cube, printed out its results then testing the instruction steps by manually manipulating the cube. This was improved until 100% reliable. I then developed the user interface to input the colours on each face. The building of the hardware to manipulate the cube proved my most difficult challenge. To get the cube flipped and rotated accurately using the 5 servos. I modeled this using lego and popsicle sticks until the movements met the accuracy and reliability outcomes I needed. Surprisingly these materials held up to the challenge. Integrating the software and hardware functional models took a lot longer than anticipated to get the software instructions executed and coordinated. A great deal of fine tuning was required. Outcome: The system solves the cube 100% of the time. I was exceptionally pleased with this result in view of the lego and popsicle stick model. On reflection I have achieved a successful working model that university students have aspired to and this gives me great satisfaction. Conclusion: While the outcome is pleasing I envisaged achieving a much faster system with easier data input using camera and colour recognition software. Unfortunately time and my budget restrictions prevented this from being developed. However this is a step I am interested in implementing in the future. The speed could be improved by designing more efficient cube solving algorithms, implementing a camera with colour recognition, and possibly rethinking and redesigning my mechanical design. I would also like to Figure 1 illustrates how the air would flow through a fan, and get pushed underground in several short HDPE pipes. This tempered air would then be fed into a small, insulated air chamber built against the home that contains an air-sourced heat pump. The walls of this chamber would have small vents to balance air pressure, and an exit near the top for cooled exhaust air. When the temperature outdoors is in the coldest stage of winter (daily average of -3.0ºC), the tempered air being brought into the chamber would simulate an outdoor ambient temperature of about 10.0ºC, allowing a heat pump to operate with a COP of ~3.79 (based on data from Goodman Air Conditioning and Heating).² This means that for every unit of energy put into the heat pump, 3.79 units of energy are extracted. 4. Conclusions: In building an enclosed air chamber for around an air-sourced heat pump, it was found that it is possible to simulate a 10.0ºC climate in the coldest parts of winter through air tempering. This will allow the heat pump to run substantially more efficiently throughout the year. This system could be used effectively to heat a home in the winter, as well as cool a home in the summer.
Developing Swarm Intelligence with Flying Robots for Life Saving
Purpose of the research I had a bicycle accident eight months ago and I suffered an opened humeral fracture. I wasn’t able to get up on my feet, I had to suffer on the ground while a couple accidentally walked there, and they called the emergency services, I lost a lot of blood and the doctors had to cut a small piece of muscle out of my biceps. Now this shouldn’t have happened if there were a faster method on making first aid. Example an intelligent flying robot could easily handle this situation. Procedures There is a hexcopter and a quadcopter. Both of them has cameras with live image transmissions in Full HD and they also have GPS. They are equipped with a lot of sensors (image recognition with Raspberry Pi, 6 axis distance sensors, long range communication modules) and with these they are able to solve problems just a bird or a human does. These drones can communicate up to 1.8 kilometers, they communicate with the main server, they share their knowledge with each other and this way they can learn about the world and about the terrain that surrounds them. In a very basic way they are able to learn… Each time they make their flying techniques smoother and their database is growing. These drones can cooperate and save lives faster than humans do. They are also able to fly in non-accessible zones by themselves with no human help. For example with their sensors they can fly in and map buildings that are inaccessible for people. They can also do artificial farming with their object recognizing features, like recognizing plants and colors. (Right now they are calibrated to look for survivors alone in the mountains/forests.) They go to a given GPS coordinate then they are flying in a spiral form to find the injured people in less than a half hour. I really hope that one day these flying intelligent machines will save lives! Note: (I probably can only take one drone with myself to Taiwan, because of the traveling restrictions on planes.)