本論文主旨在以非同步管線式(pipelined)硬體電路實現具線上學習(On-line learning)功能之類神經模糊系統。論文主題包括類神經模糊演算法的推導、其資料流的分析、整體網路硬體模組的設計及建構智慧型駕車系統模擬平台做為硬體驗證與改進參考。

　　我們所實現的類神經模糊網路架構是雙輸入、77歸屬函數的網路架構，主要硬體設計重點在以最少的硬體資源發揮高的運算效能，從演算法的化簡及演算法中儘量避開乘除法器的運算，使各硬體模組在無資料相依問題時，可達到平行或同時運算的效能。最後我們只需使用兩個乘法器及一個除法器並充分共用這些硬體資源，以FPGA進行硬體驗證達到相當高的效能(operation clock rate: 88.3 MHz, feedforward throughput: 62.9KHz, backpropagation throughput: 63.8 KHz)。

　　所設計的硬體電路在操作方面上相當容易且具有相當高的彈性，實際應用時，可隨時透過輸入誤差值，而自動進行參數調整以改善整體系統效能，完全不需額外的控制訊號設定。論文最後，我們以智慧型駕車模擬平台來驗證適應性類神經模糊控制器的正確性，經由軟體模擬與硬體驗證的結果，證實本論文所提出之具線上學習功能之類神經模糊網路硬體電路擁有高計算效能及高精確度。

　　The main focus of this thesis aims on the asynchronous pipelined hardware design of a neuro-fuzzy network with online learning capability. The topics include the derivation of neuro-fuzzy algorithm and its data-flow analysis, hardware design and implementation on an FPGA device. Finally we construct a simulation platform of an intelligent car-driving system for the verification of the effectiveness and accuracy of the overall hardware system.

　　The neuro-fuzzy network is a two-input-and-single-output system and each input is partitioned into seven term sets. That is, the fuzzy rule base consists of 77 rules. The objective of our hardware design is to utilize minimal hardware resource with maximum overall performance. To achieve this goal, we minimize the usage of multipliers and dividers, and enable hardware modules with parallel processing capability. The proposed neuro-fuzzy hardware device only requires two multipliers and one divider and its good performance includes: operation clock rate: 88.3 MHz, feedforward throughput: 62.9KHz, and backpropagation throughput: 63.8 KHz.

　　The operation of the proposed hardware device is easy and flexible. For real-world applications, the hardware device can automatically improve its performance by tuning its parameters using the error signals of the system outputs without any complex setting of control signals. The simulation results of the intelligent car-driving system indicate that the proposed neuro-fuzzy hardware device is capable of online learning desired trajectories with high computing efficiency and accuracy.

[1] T. C. Ahn, Y. W. Kwon, H. S. Hwang, and W. Pedrycz, “Design of neuro-fuzzy controller on dsp for real-time control of induction motors,” IFSA World Congress and 20th NAFIPS Int’l Conf., vol. 5, pp. 3038-3043, July 2001.
[2] H. Eichfeld, T. Kunemund, and M. Menke, “A 12b general-purpose fuzzy logic controller chip,” IEEE Trans. on Fuzzy Systems, vol. 4, no. 4, pp. 460-475, Nov. 1996.
[3] Dai. Kim and I. H. Cho, “An accurate and cost-effective fuzzy logic controller with a fast searching of moment equilibrium point,” IEEE Trans. on Industrial Electronics, vol. 46, no. 2, pp. 452-465, April 1999.
[4] C. C. Lee, “Fuzzy logic in control system: fuzzy logic controller,” IEEE Trans. on Systems, Man and Cybernetics, vol. 20, no. 2, pp. 403-435, April 1990.
[5] H. Hikawa, “Frequency-based multilayer neural network with on-chip learning and enhanced neuron characteristics,” IEEE Trans. on Neural Networks, vol. 10, no. 3, pp. 545-553, May 1999.
[6] D. Kim, “An implementation of fuzzy logic controller on the reconfigurable FPGA,” IEEE Trans. on Industrial Electronics, vol. 47, pp. 703-715, June 2000.
[7] L. M. Reyneri, “Implementation issues of neuro-fuzzy Hardware: going toward HW/SW codesign,” IEEE Trans. on Neural Networks, vol. 14, no. 1, pp. 176-194, 2003.
[8] M. Chiaberge, E. Miranda, and L. M. Reyneri, “An HW/SW co-design approach for neuro-fuzzy hardware design,” in Proceedings of the Seventh Int’l Conf. on Microelectronics for Neural, Fuzzy and Bio-Inspired Systems, pp. 332-337, April 1999.
[9] B. Bona, S. Carabelli, M. Chiaberge, E. Miranda, and L. M. Reyneri, “Hybrid neuro-fuzzy system for control of complex plants,” in Proceedings of IEEE Int’l Symposium on Industrial Electronics, vol. 1, pp. 87-92, July 1998.
[10] M. Chiaberge, E. Miranda, and L. M. Reyneri, “A hybrid digital neuro-fuzzy board for control of complex systems,” in Proceedings of IEEE Int’l Conf. on Systems, Man and Cybernetics, vol. 2, pp. 1904-1909, Oct. 1998.
[11] M. Chiaberge and L. M. Reyneri, “Cintia: a neuro-fuzzy real-time controller for low-power embedded systems,” IEEE Trans. on Micro, vol. 15, no. 3, pp. 40-47, June 1995.
[12] M. Chiaberge, E. M. Sologuren, and L. M. Reyneri, “A pulse stream system for low power neuro-fuzzy computation,” IEEE Trans. on Circuits and Systems, vol. 42, no. 11, pp. 946-954, Nov. 1995.
[13] 蘇木春、張孝德，類神經網路、模糊系統以及基因演算法則，第二版，全華，台北，民88。
[14] R. Coggins, M. Jabri, B. Flower, and S. Pickard, “A low power network for on-line diagnosis of heart patients,” IEEE Micro, vol. 15, pp. 18-25, June 1995.
[15] K. A. Boahen, “Point-to-point connectivity between neuromorphic chips using address-events,” IEEE Trans. on Circuits Syst. II, vol. 47, pp. 416-434, May 2000.
[16] S. Vassiliadis, M. Zhang, and J. G. Delgado-Frias, “Elementary function generators for neural-network emulators,” IEEE Trans. on Neural Networks, vol. 11, pp. 1438-1449, Nov. 2000.
[17] S. Jones, R. Meddis, S. C. Lim, and A. R. Temple, “Toward a digital neuromorphic pitch extraction system,” IEEE Trans. on Neural Networks, vol. 11, pp. 979-987, July 2000.
[18] M. J. Wilcox and D. C. Thelen, “A retina with parallel input and pulsed output, extracting high-resolution information,” IEEE Trans. on Neural Networks, vol. 10, pp. 574-583, May 1999.
[19] J. S. Ker, Y. H. Kuo, R. C. Wen, and B. D. Liu, “Hardware implementation of CMAC neural network with reduced storage requirement,” IEEE Trans. on Neural Networks, vol. 8, pp. 1545-1556, Nov. 1997.
[20] P. Iene, P. Thiran, and N. Vassilas, “Modified self-organizing feature map algorithms for efficient digital hardware implementation,” IEEE Trans. on Neural Networks, vol. 8, pp. 315-330, Mar. 1997.
[21] S. M. R. Hasan and N. K. Siong, “A parallel processing VLSI BAM engine,” IEEE Trans. on Neural Networks, vol. 8, pp. 424-436, Mar. 1997.
[22] G. Indiveri, “Neuromorphic analog VLSI sensor for visual tracking: Circuits and application examples,” IEEE Trans. on Circuits Syst. II, vol. 46, pp. 1337-1347, Nov. 1999.
[23] D. E. Rumelhart, J. L. McClelland, and the PDP Research Group, Parallel distributed processing: Explorations in the microstructure of cognition, MIT Press, Cambridge, MA, 1986.
[24] S. Bakshi and D. D. Gajski, “Partitioning and pipelining for performance-constrained hardware/software systems,” IEEE Trans. on VLSI Syst., vol. 7, Dec. 1999.
[25] L.-F. Chao, A. S. LaPaugh, and E. H.-M. Sha, “Rotation scheduling: A loop pipelining algorithm,” IEEE Trans. on Computer-Aided Design, vol. 16, Mar. 1997.
[26] Y.-N. Chang and K. K. Parhi, “An efficient pipelined FFT architecture,” IEEE Trans. on Circuits Syst., vol. 50, pp. 322-325, June 2003.
[27] J.-C. Latombe, Robot Motion Planning, Kluwer Academic, 1991.
[28] J. A. Reeds and R. A. Sheep, “Optimal path for a car that goes both forward and backward, ” Pacific J. Marth., vol. 145, no. 2, pp. 367-393, 1990.
[29] D.Gu and H. Hu, “Neural predictive control for car-like mobile robot,” Int’l J. of Robot. and Autonomous Systems, vol. 39, no. 2-3, May 2002.
[30] J. P. Laumond, Paul E. Jacobs, Michei Taix, and Richard M. Murry, “A motion planner for nonholonomic mobile robots,” IEEE Trans. on Robot. Automat., vol. 10, no. 5, Oct. 1994.
[31] T. Kito, J. Ota, R. Katsuki, T. Mizuta, T. Arai, T. Ueyama, and T.Nishiyama, “Smooth path planning by using visibility graph-like method,” in Proceedings of IEEE Int’l Conf. Robot. Automat., vol. 3, pp. 3770-3775, Sep. 2003.
[32] M. -M. Tomas, “Optimal path planning for car-like vehicles in the presence of obstacles,” in Proceedings of IEEE Int’l Conf. Robot. Automat., pp. 1730-1735, May 1990.
[33] F. Lamiraux and J. P. Laumond, “Smooth motion planning for car-like vehicles,” IEEE Trans. Robot. Automat., vol. 17, no. 4, pp. 498-502, Aug. 2001.
[34] T. Fraichard and J. M. Ahuactzin, “Smooth path planning for cars,” in Proceedings of IEEE Int’l Conf. Robot. Automat., pp. 21-26, Seoul, May 2001.