現今機器人於各大產業的應用非常多，而對於移動機器人而言，如何從起點移動到終點就成為了一個相當重要的一個課題，而現在，最廣泛應用於移動機器人上的便是同步定位與建置(Simultaneous Localization and Mapping, SLAM)技術，透過光達取得點雲便可即時取得周遭地圖且得知機器人於地圖的相對位置，但技術仍存在著一些缺點，在特定環境不確定性下演算法會造成明顯誤差，使得建圖發散，例如在太過寬敞的地方時，因光達能夠掃到距離不夠，而就算掃得到牆壁也會因為點雲較疏散而導致建立出來的地圖與現實不符。又或者是遇到長廊地形時，因為行經路上的特徵點都不明顯，所以就算前進很多時，SLAM演算法也會覺得沒什麼前進，甚至是沒有前進，而大大地低估了前進量。亦或者是遇到轉角時，因為轉角之幾何外型的關係，能夠光達能夠掃到的牆壁面積相較於一般時刻會大幅減少，因此轉角也是SLAM演算法會發生問題的幾種情況之一。基於上述種種原因，本研究希望能透過其它額外的感測器，給予SLAM演算法額外的資訊，令SLAM演算法在上述幾種情形下也不會失效，甚至可以降低SLAM的更新頻率需求，好讓硬體的規格降低，以達到節省成本的終極目標。而本研究所使用的感測器有旋轉編碼器和慣性量測單位中的加速規和陀螺儀，然而，不同的感測器常常可以量測或計算出相同狀態，例如陀螺儀可以量測出移動載具之航向角速度，隨著車輪旋轉的旋轉編碼器透過機器人運動學也可以計算出移動載具的航向角速度，至於兩種感測器所計算出來的角速度，要相信哪一個比較多，這個比例如何決定，即為本文之研究重點，因此本研究將透過卡爾曼濾波器來決定一個最佳比例，以估測出一個最準的狀態，並且透過自適應卡爾曼濾波器即時調整該比例，再結合本論文所提出的打滑偵測演算法，在打滑時特別調整該比例，以至於最後積分出來的路徑不至於被打滑影響太多，最後再透過模擬和實驗來驗證，該演算法確實能帶來更高的準確性。

Nowadays, robots are widely used in several industries. As far as mobile robots are concerned, how mobile robots go to the goal from the origin becomes a quite important issue. Nowadays, the technique that is most widely used for mobile robots is Simultaneous Localization And Mapping (SLAM). SLAM can build the ambient map online and localize the relative position through the point cloud scanned from LiDAR. But there are still some shortcomings in this technology, the algorithm will go wrong in certain environment conditions, such as, out of scan limit, cloister(not enough characteristic feature). The vision of this research is to improve the above problems by combining other sensors which can provide SLAM algorithm additional information to make SLAM algorithm doesn’t go wrong in mentioned situations. Even the pre-compensation mechanism can help SLAM to turn the update frequency down and downgrade the hardware so that the final goal of saving budget can be expected. The sensors which are used in this research are encoders, gyros and accelerometers. However, the data measured by different sensors often can be used to calculate same state. For example, the data measured from gyro and encoder both can be used to calculate the yawing rate of Wheeled Mobile Robot (WMR). As for which one should be trusted more. How to decide the weighting of trust is also the main purpose of this research. Therefore, this research will decide an optimal weighting through Kalman Filter (KF) to estimate the optimal state. After that, this research will combine Adaptive Kalman Filter (AKF) and proposed slippage detection algorithm to decide the weighting when slippage happens so that the estimated posture of WMR will not be influence by the effect of slippage. At last, this research will confirm the theory through simulation and experiment.

[1] S. G. Tzafestas, Introduction to mobile robot control. Elsevier, 2013.
[2] c. Wikipedia. (26 July 2020 11:10 UTC). Kalman filter. Available: https://en.wikipedia.org/w/index.php?title=Kalman_filter&oldid=967968225
[3] G. Reina, G. Ishigami, K. Nagatani, and K. Yoshida, "Vision-based estimation of slip angle for mobile robots and planetary rovers," in 2008 IEEE International Conference on Robotics and Automation, 2008, pp. 486-491: IEEE.
[4] N. K. Goswami and P. K. Padhy, "Gain tuning of Lyapunov function based controller using PSO for mobile robot control," in 2016 11th International Conference on Industrial and Information Systems (ICIIS), 2016, pp. 295-299: IEEE.
[5] C. C. Ward and K. Iagnemma, "Model-based wheel slip detection for outdoor mobile robots," in Proceedings 2007 IEEE International Conference on Robotics and Automation, 2007, pp. 2724-2729: IEEE.
[6] C. C. Ward and K. J. I. T. o. R. Iagnemma, "A dynamic-model-based wheel slip detector for mobile robots on outdoor terrain," vol. 24, no. 4, pp. 821-831, 2008.
[7] G. D'Agostini, "A multidimensional unfolding method based on Bayes' theorem," P000243781994.
[8] J. Yi, J. Zhang, D. Song, and S. Jayasuriya, "IMU-based localization and slip estimation for skid-steered mobile robots," in 2007 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2007, pp. 2845-2850: IEEE.
[9] G. Welch and G. Bishop, "An introduction to the Kalman filter," 1995.
[10] M. Hoshiya and E. J. J. o. e. m. Saito, "Structural identification by extended Kalman filter," vol. 110, no. 12, pp. 1757-1770, 1984.
[11] S. J. Julier and J. K. J. P. o. t. I. Uhlmann, "Unscented filtering and nonlinear estimation," vol. 92, no. 3, pp. 401-422, 2004.
[12] S. J. Julier and J. K. Uhlmann, "New extension of the Kalman filter to nonlinear systems," in Signal processing, sensor fusion, and target recognition VI, 1997, vol. 3068, pp. 182-193: International Society for Optics and Photonics.
[13] R. Van Der Merwe, A. Doucet, N. De Freitas, and E. A. Wan, "The unscented particle filter," in Advances in neural information processing systems, 2001, pp. 584-590.
[14] M. NøRgaard, N. K. Poulsen, and O. J. A. Ravn, "New developments in state estimation for nonlinear systems," vol. 36, no. 11, pp. 1627-1638, 2000.
[15] R. Van Der Merwe, E. Wan, and S. Julier, "Sigma-point Kalman filters for nonlinear estimation and sensor-fusion: Applications to integrated navigation," in AIAA Guidance, Navigation, and Control Conference and Exhibit, 2004, p. 5120.
[16] J. J. Moré, "The Levenberg-Marquardt algorithm: implementation and theory," in Numerical analysis: Springer, 1978, pp. 105-116.
[17] A. Alamin, H. M. Khalid, and J. C.-H. Peng, "Power system state estimation based on Iterative Extended Kalman Filtering and bad data detection using normalized residual test," in 2015 IEEE Power and Energy Conference at Illinois (PECI), 2015, pp. 1-5: IEEE.
[18] R. Zhan, J. J. I. T. o. A. Wan, and E. Systems, "Iterated unscented Kalman filter for passive target tracking," vol. 43, no. 3, pp. 1155-1163, 2007.
[19] G. Sibley, G. S. Sukhatme, and L. H. Matthies, "The iterated sigma point kalman filter with applications to long range stereo," in Robotics: Science and Systems, 2006, vol. 8, no. 1, pp. 235-244.
[20] Y. Wu, Z. Xia, and X. Liang, "Battery SOC Estimation Based on LM-ICDKF Algorithm," in 2017 10th International Conference on Intelligent Computation Technology and Automation (ICICTA), 2017, pp. 25-28: IEEE.
[21] 刘学 and 焦. J. 深圳大学学报理工版, "基于迭代 CDKF 的单站无源定位算法," vol. 28, no. 2, 2011.
[22] 杨宏, 李亚安, 李国辉, and 袁润平, "一种改进中心差分卡尔曼滤波方法," 2010.
[23] I. Hashlamon, K. J. T. j. o. e. e. Erbatur, and c. sciences, "An improved real-time adaptive Kalman filter with recursive noise covariance updating rules," vol. 24, no. 2, pp. 524-540, 2016.
[24] 姜浩楠 and 蔡. J. 控制与决策, "带有噪声递推估计的自适应集合卡尔曼滤波," vol. 33, no. 9, pp. 1567-1574, 2018.
[25] Y. Shi, C. Han, and Y. Liang, "Adaptive UKF for target tracking with unknown process noise statistics," in 2009 12th International Conference on Information Fusion, 2009, pp. 1815-1820: IEEE.
[26] Z. Gao, D. Mu, S. Gao, Y. Zhong, C. J. A. S. Gu, and Technology, "Adaptive unscented Kalman filter based on maximum posterior and random weighting," vol. 71, pp. 12-24, 2017.
[27] B. Zheng, P. Fu, B. Li, and X. J. S. Yuan, "A robust adaptive unscented Kalman filter for nonlinear estimation with uncertain noise covariance," vol. 18, no. 3, p. 808, 2018.
[28] J. Xu, Y. Jing, G. M. Dimirovski, and Y. Ban, "Two-stage unscented Kalman filter for nonlinear systems in the presence of unknown random bias," in 2008 American Control Conference, 2008, pp. 3530-3535: IEEE.
[29] H. Wang, G. Fu, J. Li, Z. Yan, and X. J. M. P. i. E. Bian, "An adaptive UKF based SLAM method for unmanned underwater vehicle," vol. 2013, 2013.
[30] F. Janabi-Sharifi and M. J. I. t. o. R. Marey, "A kalman-filter-based method for pose estimation in visual servoing," vol. 26, no. 5, pp. 939-947, 2010.
[31] V. Lippiello, B. Siciliano, and L. J. C. E. P. Villani, "Adaptive extended Kalman filtering for visual motion estimation of 3D objects," vol. 15, no. 1, pp. 123-134, 2007.
[32] J. Hammersley, Monte carlo methods. Springer Science & Business Media, 2013.
[33] S.-S. Jan and Y.-C. J. S. Kao, "Radar tracking with an interacting multiple model and probabilistic data association filter for civil aviation applications," vol. 13, no. 5, pp. 6636-6650, 2013.