本論文首先設計群集機器人架構後使用3D印表機列印機器人輪胎，接著設計電路完成焊接並組裝，再來使用藍牙與超音波來完成環境建構跟路徑規劃，分別由六台搭載藍牙以及超音波模組的輪型機器人，利用藍牙將六台輪型機器人同步連接到開發板Raspberry Pi後利用超音波得出環境資料傳回。開發板Raspberry Pi將收集的資料進行環境的建構得出當下的環境，接著使用Q學習演算法對環境進行路徑規劃得出當下每台機器人對環境的移動路徑傳回輪型機器人，如果輪型機器人在移動中在原本要行走的路徑上突然出現障礙物，則會立刻停下並回傳障礙物資料後重新執行演算法進行路徑規劃，演算法運算的過程輪型機器人是處於停下的狀態，如果通訊的速度跟演算法運算的速度很快的話，能使輪型機器人較平滑移動後進入下一條路徑移動，讓整體移動的更為流暢，最後六台輪型機器人都會依照自己跟環境所運算得到的路徑去移動通過障礙物。

In this paper we use Q-learning algorithm performed on the Raspberry Pi to construct to path plan for swarm robots. We first designs the swarm robot archi-tecture and uses the 3D printer to produce the tires of robots. Then we finish the design, welding and assembly of the circuit. The path planning and environ-ment building is completed by using Bluetooth and Ultrasound sensor in the fol-lowing manner: six mobile robots carry Bluetooth to connect synchronically to the Raspberry Pi. Then the robots use Ultrasound to detect the environment. The data of environment are then transferred to the Raspberry Pi by using Bluetooth so that the model of environment can be constructed by the Raspberry Pi.To find the optimal path for the robots, Q-learning algorithm is employed in the Raspberry Pi to calculate the track of every robot at every moment. Calculated results are then transferred to the robots for the next movement. In case that there are obstacles present at the planned path for robots, the robots will stop and transfer the infor-mation of obstacles and the path plan will be renewed by using Q-learning algo-rithm.During the calculation, the robots stay at their present position until the paths are found for next movement. Therefore, the movements of robots will be more smoothly if the speed of data transfer and calculation of path are high enough because the robots can move to next position without significant stay.As a result, the robots can pass the obstacles according to path plan.

[ 1 ]Doan K.N., Le A.T., Le T.D., Peter N. "Swarm Robots’ Communication and Cooperation in Motion Planning." Mechatronics and Robotics Engineering for Advanced and Intelligent Manufac-turing. Springer International Publishing, 2017. 191-205.
[ 2 ]Wurm, Kai M., Cyrill Stachniss, and Wolfram Burgard. "Coordinated multi-robot exploration using a segmentation of the environment." Intelligent Robots and Systems, 2008. IROS 2008. IEEE/RSJ International Conference on. IEEE, 2008.
[ 3 ]Brutschy, Arne, Lorenzo Garattoni,Manuele Brambilla,Gianpiero Francesca,Giovanni Pini,Marco Dorigo,Mauro Birattari. "The TAM: abstracting complex tasks in swarm robotics research." Swarm Intelligence 9.1 （2015）: 1-22.
[ 4 ] M.A. PortaGarcia,Oscar Montiel,Oscar Castillo,Roberto Sepúlveda,Patricia Melin."Path planning for autonomous mobile robot navigation with ant colony optimization and fuzzy cost function evaluation." Applied Soft Computing 9.3 （2009）: 1102-1110.
[ 5 ] Micael S.Couceiro,Patricia A.Vargas,Rui P.Rocha,Nuno M.F.Ferreira. "Benchmark of swarm robotics distributed techniques in a search task." Robotics and Autonomous Systems 62.2 （2014）: 200-213.
[ 6 ] Kaelbling, Leslie Pack, Michael L. Littman, and Andrew W. Moore. "Reinforcement learning: A survey." Journal of artificial intelligence research 4 （1996）: 237-285.
[ 7 ]汪昌賢,類神經網路自組織增強式學習模型,國立台灣大學電機資訊學院資訊工程學系,2016
[ 8 ] Cassandra, Anthony Rocco. "Exact and approximate algorithms for partially observable Markov decision processes." （1998）.
[ 9 ] Lewis, Frank L., and Draguna Vrabie. "Reinforcement learning and adaptive dynamic pro-gramming for feedback control." IEEE circuits and systems magazine 9.3 （2009）.
[ 10 ] Bradtke, Steven J., and Andrew G. Barto. "Linear least-squares algorithms for temporal dif-ference learning." Machine learning 22.1-3 （1996）: 33-57.
[ 11 ] White, Martha, and Adam White. "A Greedy Approach to Adapting the Trace Parameter for Temporal Difference Learning." Proceedings of the 2016 International Conference on Autonomous Agents & Multiagent Systems. International Foundation for Autonomous Agents and Multiagent Systems, 2016.
[ 12 ] Tijsma, Arryon D., Madalina M. Drugan, and Marco A. Wiering. "Comparing exploration strategies for Q-learning in random stochastic mazes." IEEE Symposium Series on
[ 13 ] CAO Wei-hua,XU Ling-yun,WU Min.”A double-layer decision-making model based on fuzzy Q-learning for robot soccer,”（2008）03-0234-05
[ 14 ] Nils Morozs,Tim Clarke, David Grace,Qiyang Zhao. "Distributed Q-learning based dynamic spectrum management in cognitive cellular systems: Choosing the right learning rate." Computers and Communication （ISCC）, 2014 IEEE Symposium on. IEEE, 2014.
[ 15 ] Cristiane Silva Garcia,Diego Eckard,Joao Cesar Netto,Carlos Eduardo Pereira,Ivan Muller. "Bluetooth Enabled Data Collector for Wireless Sensor Networks." Computing Systems Engineering （SBESC）, 2015 Brazilian Symposium on. IEEE, 2015.
[ 16 ] Mandal, Bijoy Kumar, Debnath Bhattacharyya, and Tai-hoon Kim. "A Design Approach for Wireless Communication Security in Bluetooth Network." International Journal of Security and Its Applications 8.2 （2014）: 341-352.
[ 17 ] Kalia, Manish, Deepak Bansal, and Rajeev Shorey. "MAC scheduling and SAR policies for Bluetooth: A master driven TDD pico-cellular wireless system." Mobile Multimedia Communications, 1999.(MoMuC'99) 1999 IEEE International Workshop on. IEEE, 1999.
[ 18 ] Ling-Jyh Chen,R. Kapoor, M.Y. Sanadid,M. Gerla. "Enhancing Bluetooth TCP throughput via link layer packet adaptation." Communications, 2004 IEEE International Conference on. Vol. 7. IEEE, 2004.
[ 19 ] Gazis, Denos C., Robert S. Jaffe, and William G. Pope. "Optimal and stable route planning system." U.S. Patent No. 5,610,821. 11 Mar. 1997.
[ 20 ] He Bing,Liu Gang,Gao Jiang,Wang Hong,Nan Nan,Li Yan. "A route planning method based on improved artificial potential field algorithm." Communication Software and Networks （ICCSN）, 2011 IEEE 3rd International Conference on. IEEE, 2011.
[ 21 ] 朱大奇、顏明重(2010)。移動機器人路徑規劃技術綜述，控制與決策，25:7， 961-967。
[ 22 ] 默凡凡, 基于Q學習算法的移動機器人路徑規劃方法研究,北京工業大學,2016
[ 23 ] Klidbary, Sajad Haghzad, Saeed Bagheri Shouraki, and Soroush Sheikhpour Kourabbaslou. "Path planning of modular robots on various terrains using Q-learning versus optimization algo-rithms." Intelligent Service Robotics 10.2 (2017): 121-136.
[ 24 ] Pratyusha Rakshit,Amit Konar,Pavel Bhowmik,Indrani Goswami,Swagatam Das,Lakhmi C. Jain,Atulya K. Nagar. "Realization of an adaptive memetic algorithm using differential evolution and q-learning: a case study in multirobot path planning." IEEE Transactions on Systems, Man, and Cybernetics: Systems 43.4 (2013): 814-831.
[ 25 ] 直流馬達的PWM調速控制,http://eshare.stust.edu.tw/EshareFile/2010_4/2010_4_ccd9befb.pdf
[ 26 ] David C. Duffy,J Cooper McDonald,Olivier J. A. Schueller,George M. Whitesides.”Rapid prototyping of microfluidic systems in poly(dimethylsiloxane).”Analytical chemistry 70.23 (1998):4974-4984
[ 27 ] Dudek, P. "FDM 3D printing technology in manufacturing composite elements." Archives of Metallurgy and Materials 58.4 （2013）: 1415-1418.
[ 28 ] Flynn, Anita M. "Combining sonar and infrared sensors for mobile robot navigation." The In-ternational Journal of Robotics Research 7.6 （1988）: 5-14.
[ 29 ] Arduino nano 設計原理,http://download.arduino.org/products/NANO/Arduino%20Nano-Rev3.2-SCH.pdf
[ 30 ]樹莓派官方使用手冊, http://www.cypress.com/file/298076/download
[ 31 ]TOSHIBA 公司,” TA7279P DataSheet”, https://www.bucek.name/pdf/ta7279.pdf
[ 32 ] MICROPIK 公司,”HC-SR04 DataSheet”, http://www.micropik.com/PDF/HCSR04.pdf
[ 33 ]ITead studio 公司,”HC-05 DataSheet”, http://www.electronicaestudio.com/docs/istd016A.pdf