• 沒有找到結果。

第四章 具有高計算效率之視覺型即時定位與建圖演算法 (V-CESLAM)

4.1 視覺地標

4.1.1 SURF 特徵比對

SURF 為了加快比對的速度,在判斷兩個特徵是否為同一特徵時,會先判斷 這兩個特徵的拉氏信號(Sign of Laplacian),也就是看這兩個特徵的 Hessian 矩陣 的行列式值是否為極大值或是極小值,而拉氏信號能夠分辨出不同的特徵點的對

接下來使用最近鄰點距離比對法(Nearest Neighbor Distance Ratio, NNDR)[46],假 設粒子中已存有N 個地標

l n

n

,( =  1 N )

,機器人從影像擷取出的特徵為

s

,利用

38

圖4.6、特徵與機器人之實際空間距離

依序找到影像I 中所有可用的特徵後,將第g i 個特徵依序對粒子

m

內所 含的

N

個地標進行SURF 比對的動作,找出此特徵是否為已知地標或需要新 增地標。

3. 粒子狀態更新、地標更新與新增:

a、 地標更新:

當影像中特徵i 為粒子

m

中的已知地標k 時,我們將沿著 CESLAM 的做法,在更新地標之前,先使用這組比對成功的數據先更新粒子的狀 態。但現在的資料已經變成三維的,而之前的 CESLAM 是採用二維的 計算方式,為了保持計算速度以達成即時的特性,我們在以下公式中皆 捨棄

y

維度,因為機器人跟地標的高度皆可假設為固定,並不會因為機 器人的移動而有所改變,因此可將圖4.6 改為俯視圖,如圖 4.7 所示。

39

並計算創新共變異數矩陣(innovation covariance):

k , 1 k

40

41

表 5.1、Kinect 規格

44

5.1.2 計算平台及軟體

本論文使用的筆記型電腦為 ACER 的 4820TG,其規格如表 5.2 所示。程式 部分則是使用Visual Studio 2010,以 C 語言撰寫,並搭配 OpenCV 函式庫,使用 其所提供的 SURF 及矩陣的相關函式,以及顯示影像相關的函式。同時也使用 OpenNI 函式庫以及微軟官方的 Kinect for Windows SDK 讀取 Kinect 的彩色影像 及深度資訊。 landmark,粒子數為 40。機器人每次移動距離為 10 pixel,每次旋轉角度為 15 度,

且里程計所回傳的移動距離與旋轉角度都帶有高斯雜訊。模擬的LRF 範圍有 180

45

色圓圈的大小沒有特別意義,只是為了方便觀察地標位置。紅色圓圈的圓心為粒 子所估測出的地標位置,藍色線段為模擬的雷射測距儀。圖 5.3 明顯地顯示出我 們所提出的 CESLAM 不管是在路徑的預測還是地標的估測,其效能都遠優於 FastSLAM2.0 跟 FastSLAM1.0。

47

面都優於FastSLAM2.0。在路徑誤差方面,CESLAM 的精準度提升了約 17%,而 估測地標的精準度也提升了20%,且成功率提升了 5%。

表 5.4 為各種 FastSLAM 演算法在不同地標數時之執行時間,當地標數目較 少時,FastSLAM 1.0 還可以比 CESLAM 快上許多,但 CESLAM 已經比 FastSLAM 2.0 來的快,而且當地標數量增加時,越能展現出 CESLAM 的優勢。而當地標數

48

圖 5.4、各種 SLAM 的執行時間

圖5.4 為三種演算法的執行時間折線圖。橫軸代表目前所執行到的步數。縱 軸為時間,單位是秒。SLAM 任務剛開始時,因為粒子中所存的地標較少,所以 三種演算法所需的時間相差不多。但隨著機器人不斷地探索環境,地標數量逐漸 增加,可看出FastSLAM 2.0 所花費的時間係以指數型快速的增加中,而 CESLAM

0

FastSLAM1.0 40.76 44.13 98.54

FastSLAM2.0 45.10 51.45 184.80

CESLAM 44.03 47.71 72.44

表5.5、直線移動之地面基準點實驗數據 Ground Truth (m) Estimated

x(m) y(m)

θ

(度) x( , )

µ σ

(m) y( , )

µ σ

(m)

θ µ σ

( , ) (度) 0.00 0.00 90°

( 0.0053,0.0268) − (0.0010,0.0185) (89.7667,0.9551)°

0.00 1.00 90°

(0.0147,0.0358) ( 0.0120,0.0245) − (90.2667,1.2892)°

表5.6、原地旋轉之地面基準點實驗數據 Ground Truth Estimated

x(m) y(m)

θ

(度) x( , )

µ σ

(m) y( , )

µ σ

(m)

θ µ σ

( , ) (度) 0.00 0.00 90°

(0.0274,0.0373) ( 0.0089,0.0533) − (86.6667,4.7889)°

0.00 0.00 180°

(0.0413,0.0245) ( 0.0267,0.0272) − (176.2667,2.112)°

0.00 0.00 0°

(0.0520,0.0489) (0.0067,0.0499) (356.4667,2.6300)°

67

標後,結果即改善很多。而保持每次旋轉 15 度的原因是為了證明機器人在大幅 度的改變影像時依然能夠定位出自己的位置,因此不改變旋轉的角度。由實驗數 據也可看出誤差已在可接受的範圍內,建立出來的地圖也跟實際環境非常相似。

為了測試V-CESLAM 是一可行的方法,在 5.4 節中我們實際使用 P-3DX 機 器人進行loop closure 的實驗,命令機器人旋轉 360 度,測試當環境經過大幅度 的改變後,回到已知點時是否能正確辨識出機器人所在的位置。而由實驗結果可 看出在機器人在繞完一圈回到原點時,能夠偵測出最一開始的地標,建圖的結果 也與實際狀況非常相似,代表V-CESLAM 是一實際可解決 SLAM 問題的方法。

接著在5.5 節中我們應用 V-CESLAM 演算法於現實狀況,讓機器人實際的在環境 中依照方形的路徑行走並建圖,其結果顯示出本方法即使機器人在經過移動與旋 轉後回到起始點形成 loop closure 時,依舊能定位出機器人實際的位置,同時並 建立出與實際環境相符的特徵地圖。

68

第六章 結論與未來展望

6.1 結論

本論文以FastSLAM 演算法為基礎,提出「具有高計算效率之及時定位與建 圖演算法(CESLAM)」演算法解決 SLAM 的問題,捨棄一開始在 FastSLAM2.0 中 利用環境資訊更新粒子位置的階段,而改成先用里程計資訊更新粒子,並只選擇

69

參考文獻

[1].

H. Durrant-Whyte and T. Bailey, “Simultaneous localization and mapping: Part I,”

IEEE Robot Autom. Mag., vol. 13, no. 2, pp. 99–108, Jun. 2006.

[2].

J.J. Leonard and H.F. Durrant-Whyte, “Mobile robot localization by tracking geometric beacons,” IEEE Transactions on Robotics and Automation, vol. 7, no.

3, pp. 376-382, 1991.

[3].

H. Durrant-Whyte, D. Rye, and E. Nebot, “Localisation of automatic guided vehicles,” The 7th International Symposium on Robotics Research (ISRR’95), 1996, pp. 613-625.

[4].

J.J. Leonard and H.J.S. Feder, “A computationally efficient method for large-scale concurrent mapping and localization,” The Ninth International Symposium on Robotics Research (ISRR’99), 2000, pp. 169-176.

[5].

J. Guivant, E. Nebot, and S. Baiker, “Localization and map building using laser range sensors in outdoor applications,” Journal of Robotic Systems, vol. 17, no.

10, pp. 565-583, 2000.

[6].

S.B. Williams, P. Newman, G. Dissanayake, and H.F. Durrant-Whyte,

“Autonomous underwater simultaneous localisation and map building,” IEEE International Conference on Robotics and Automation (ICRA), San Francisco, 2000, vol. 2, pp. 1793-1798.

[7].

F. Lu and E. Milios, “Robot pose estimation in unknown environments by matching 2D range scans,” Journal of Intelligent and Robotic Systems: Theory and Applications, vol. 18, no. 3, pp. 249-275, 1997.

[8].

T. Duckett, S. Marsland, and J. Shapiro, “Fast, on-line learning of globally consistent maps,” Autonomous Robots, vol. 12, no. 3, pp. 287-300, 2002.

[9].

K. Pathak, M. Pfingsthorn, N. Vaskevicius, and A. Birk, “Relaxing loop-closing errors in 3D maps based on planar surface patches,” 2009 International Conference on Advanced Robotics (ICAR 2009), Munich, Germany, 2009, pp.

1-6.

[10].

K. Pathak, N. Vaskevicius, and A. Birk, “Uncertainty analysis for optimum plane extraction from noisy 3D range-sensor point-clouds,” Intelligent Service Robotics, vol. 3, no. 1, pp. 37-48, 2009.

70

[11].

F. Masson, J. Guivant, and E. Nebot, “Hybrid architecture for simultaneous localization and map building in large outdoor areas,” IEEE International Conference on Intelligent Robots and Systems, vol. 1, pp. 570-575, 2002.

[12].

J.E. Guivant, F.R. Masson, and E.M. Nebot, “Simultaneous localization and map building using natural features and absolute information,” Robotics and Autonomous Systems, vol. 40, no. 2-3, pp. 79-90, 2002.

[13].

R.C. Smith and P. Cheeseman, “On the representation and estimation of spatial uncertainty,” International Journal of Robotics, vol. 5, no. 4, pp. 56-58, 1986.

[14].

S.B. Williams, G. Dissanayake, and H.F. Durrant-Whyte, “An efficient approach to the simultaneous localisation and mapping problem,” IEEE International Conference on Robotics and Automation, Washington, USA, 2002, vol. 1, pp.

406-411.

[15].

G. Dissanayake, S.B. Williams, H.F. Durrant-Whyte, and T. Bailey, “Map management for efficient simultaneous localization and mapping (SLAM),”

Autonomous Robots, vol. 12, no. 3, pp. 267-286, 2002.

[16].

A. Chatterjee, “Differential evolution tuned fuzzy supervisor adapted, extended Kalman filtering for SLAM problems in mobile robots,” Robotica, vol. 27, pp.

411-423, 2009.

[17].

A. Chatterjee and F. Matsuno, “A neuro-fuzzy assisted extended Kalman filter based approach for Simultaneous Localization and Mapping (SLAM) problems,”

IEEE Transactions on Fuzzy Systems, vol. 15, pp. 984-997, 2007.

[18].

A. Chatterjee and F. Matsuno, “A Geese PSO tuned fuzzy supervisor for EKF based solutions of simultaneous localization and mapping (SLAM) problems in mobile robots,” Expert Systems with Applications, vol. 37, pp. 5542-5548, 2010.

[19].

A. Stentz, D. Fox and M. Montemerlo, “FastSLAM: A factored solution to the simultaneous localization and mapping problem with unknown data association,”

In Proceedings of the AAAI National Conference on Artificial Intelligence, 2003, pp. 593-598.

[20].

G. Dissanayake, P. Newman, H.F. Durrant-Whyte, S. Clark, and M. Csobra, “An experimental and theoretical into simultaneous localization and map building (SLAM),” Lecture Notes in Control and Information Sciences, Experimental

71

Robotics VI, 2000, pp. 265-274.

[21].

K. Murphy, “Bayesian map learning in dynamic environments,” Neural Information Proceedings System, vol. 12, pp. 1015–1021, 2000.

[22].

M. Montemerlo, S. Thrun, D. Koller, and B. Wegbreit, “FastSLAM 2.0: An improved particle filtering algorithm for simultaneous localization and mapping that provably converges,” International Joint Conference on Artificial Intelligence, 2003, pp. 1151-1156.

[23].

M. Montemerlo and S. Thrun, “Simultaneous localization and mapping with unknown data association using FastSLAM,” Proc. IEEE Int. Conf. Robotics and Automation, 2003, pp. 1985–1991.

[24].

鄧宏志,結合粒子群最佳化法之雙層粒子濾波器於移動機器人的定位與地圖

建置,博士論文,淡江大學電機工程學系,民國100 年。

[25].

陳雨政,分離更新式FastSLAM 之設計與實現,博士論文,機械與機電工程

學系,民國101 年。

[26].

A. J. Dacison, I. D. Reid, N. Molton, and O. Stasse, “Monoslam: Real-time single camera slam, ” IEEE transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 6, pp. 1052-1067, June 2007.

[27].

L. M. Paz, P. Pinies, J. D. Tardos and J. Neira, “Large-Scale 6-DOF SLAM with Stereo-in-Hand, ” IEEE transactions on Robotics and Automation, vol. 24, no. 5, pp. 946-957, 2008.

[28].

P. Henry, M. Krainin, E. Herbst, X. Ren, and D. Fox, “RGB-D Mapping: Using depth cameras for dense 3D modeling of indoor environments,” in the 12th International Symposium on Experimental Robotics (ISER), 2010.

[29].

T. Emter, A. Stein, “Simultaneous Localization and Mapping with the Kinect sensor,” Proceedings of ROBOTIK, Munich, Germany, 2012, pp. 239-244.

[30].

J. Hartmann, D. Forouher, M. Litza, J. H. Klüssendorff and E. Maehle,

“Real-Time Visual SLAM Using FastSLAM and the Microsoft Kinect Camera,”

Proceedings of ROBOTIK, Munich, Germany, 2012, pp. 458-463.

[31].

Y. C. Ho and R. Lee, “A Bayesian approach to problems in stochastic estimation and control,” IEEE Transaction on Automatic Control, 2001, vol. AC-9, pp.

72

333-339.

[32].

S. J. Julier and J. K. Uhlmann “A New Extension of the Kalman Filter to Nonlinear Systems,” Proc. of AeroSense: The 11th Int. Symp. on Aerospace/Defence Sensing, Simulation and Controls, Orlando, USA, 1997, vol.

3068, pp. 182-193.

[33].

M.S. Arulampalam, S. Maskell, N. Gordon, and T. Clapp, “A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking,” IEEE Transactions on Signal Processing, vol. 50, no. 2, pp. 174-188, 2002.

[34].

G. Kitagawa, “Monte Carlo filter and smoother for non-Gaussian nonlinear state space models,” Journal of Computational and Graphical Statistics, vol. 5, no. 1, pp. 1-25, 1996.

[35].

F. Dellaert, D. Fox, W. Burgard, and S. Thrun, “Monte Carlo localization for mobile robots,” IEEE International Conference on Robotics and Automation, 1999, pp. 1322-1328.

[36].

S. Thrun, W. Burgard, and D. Fox, Probabilistic Robotics, the MIT Press, 2005.

[37].

I. Rekleitis. “A Particle Filter Tutorial for Mobile Robot Localization,” Technical Report TR-CIM-04-02, Centre for Intelligent Machines, McGill University, Montreal, Quebec, Canada, 2004.

[38].

Chen-Chien Hsu, Ching-Chang Wong, Hung-Chih Teng, and Cheng-Yao Ho,

“Localization of Mobile Robots via an Enhanced Particle Filter Incorporating Tournament Selection and Nelder-Mead Simplex Search,” International Journal of Innovative Computing, Information and Control, vol. 7, no. 7A, pp. 3725-3737, July, 2011.

[39].

H. Bay, T. Tuytelaars, and L. Van Gool, “SURF: Speeded up robust features,”

Proceedings of European Conference on Computer Vision, 2006, pp. 404-417.

[40].

D. G. Lowe, “Object recognition from local scale-invariant,” International Conference of Computer Vision, 2004, vol. 60, pp. 1150-1157.

[41].

P. Viola and M. Jones, “Rapid object detection using a boosted cascade of simple features,” Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003, vol. 2, pp.

1403-1410.

73

[42].

D. Lowe, “Distinctive image features from scale-invariant keypoints,”

International Journal of Computer Vision, vol.60, no.2, pp.91-110, 2004.

[43].

R. Lienhart and J. Maydt, “An Extended Set of Haar-Like Features for Rapid Object Detection,” Proc. IEEE Int’l Conf. Image Processing, 2002, vol. 1, pp.

900-903.

[44].

H. P. Moravec, “Towards Automatic Visual Obstacle Avoidance,” Proc. 5th International Joint Conference on Artificial Intelligence, 1977, pp. 584.

[45].

C. Harris and M. Stephens, “A Combined Corner and Edge Detector,”

Proceedings of Alvey Vision Conference, Manchedter, 1988, pp. 147-151.

[46].

K. Mikolajczyk and C. Schmid, “A Performance Evaluation of Local Descriptors,”

IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 27, no. 10, pp. 1615-1630, 2005.

[47].

http://www.xbox.com/zh-TW/Kinect

74

75

許老師如此用心的栽培我。

為了比別人多一份競爭力,在研究所的第一年中,我不斷的修習各種領域的 課程,像是數位相機設計、影像處理、數值分析等,且都保持著不錯的成績,碩 一上是全班第一名,碩一下則是第二名。同時不斷的精進自己的各種程式能力,

像是C、C++、Verilog 等,並不斷的學習英文,希望能讓自己符合社會未來的趨 勢。

3、 個人經驗

在大學中,學業固然重要,但不是唯一。在這個小型社會裡,與師長、同學 間的相處、對事情的看法,也是我們必須要增進的部分,於是我參加了系學會(對 內)與吉他社(對外)。大二時,承蒙同學們的支持,我成為了應電系系學會長,

同時也成為了吉他社的幹部。當會長時,與幹部們的溝通、勇於決定一切事物,

並承擔後果,都是他人無法學習到的經驗。當吉他社幹部時,勇於表達自己的意 見,並虛心接受學長姐的指導,讓我回到系上當會長時能對幹部們有同理心。這 一年雖然累,但累得很滿足。 大三時,成為了吉他社的教學組理事。因為大二 時的經驗較為豐富,所以我對於活動的看法總是比他人更加細膩,並在寒假時主 辦了人數近100 人的師大全國吉他營,從計畫、執行、到檢討,每一件事都讓我 的想法與經驗更加成熟,同時也增進了與講師、學員間的溝通能力。大四則當上 吉他社的監事長,把我這二年來擔任幹部、系學會會長、教學組理事的經驗,毫 不自私的全部傳承給下一屆,讓學弟妹能做青出於藍,更勝於藍。

76

學術成就

1. 論文發表

C. K. Yang, C. C. Hsu, Y. T. Wang, “Computationally Efficient Algorithm for Simultaneous Localization and Mapping (SLAM), ” IEEE International Conference on Networking, Sensing and Control(ICNSC2013), France, April, 2013, pp. 328-332.

2. 專利

中華民國專利 「移動物體軌跡偵測系統(Trajectory detection system for moving object)」 (審查中)。

3. 競賽

2011 亞洲創新設計大賽「不球於人(棒球軌跡偵測系統)」佳作獎。

4. 參與研究計畫

99B0332-以 SOPC 為基礎利用單一攝影機傾斜攝影之移動式機器人物體追 蹤與定位系統。

99B0452-02-應用於室內自然環境且可與人自然溝通之居家服務機器人-子計

99B0452-02-應用於室內自然環境且可與人自然溝通之居家服務機器人-子計

相關文件