橫向攀爬機器人所應用之壁面環境具有複雜特徵，每種壁面特徵需要相對應的攀爬步態才得以抓握，使得辨識環境變化與自主修正步態的能力至關重要；既有攀爬機器人辨識環境能力的研究中，由於視覺感測器裝設視角固定的關係，僅能辨識少量的環境特徵，且無法適應具自由空間（Free Space）之環境。所以，本研究提出一種以接觸力資訊進行環境探索與路徑規劃的演算法，如同盲人沿著牆壁邊摸邊前進一樣，使橫向攀爬機器人能夠適應具連續變化與自由空間的攀爬環境。在接觸力回授系統的設計上，本研究採用動量估測器(Momentum Observer)量測機器人與環境間的碰撞力量；並且，透過時變閥值(Time Varying Threshold)的訂定，區分系統所受之干擾為移動慣性或碰撞力，並在機器人與環境發生碰撞時，以導納控制律(Admittance Law)計算移動修正量，切換移動軌跡，以適應具碰撞之攀爬情況。此外，當機器人攀爬至自由空間(Free Space)的環境時，基於A*演算法的搜索環境概念，以移動距離當作運動規劃考量，對當前活動空間進行環境探索，搜尋空間中的可抓握點。最後，以二連桿機器人當作系統模型，並且以模擬方式成功驗證本研究設計的基於接觸力資訊之環境探索與運動規劃演算法，可以使橫向攀爬機器人適應具：水平、水平不連續、傾斜、高低差(水平)、高低差(傾斜)的攀爬環境。

Transverse climbing robots are applied to the tasks involving transverse movements along the complex features on the wall, such as horizontal and inclined ledges, and gap between ledges with different elevations. The abilities of recognizing wall feature variations and correctly switching locomotion strategies to adapt to the environment changes are very crucial for safe autonomous climbing. However, the methods developed with the widely used visual sensors are all suffering from the occlusion problem and thus cannot be fully applied to exploring free space environments due to the limited information of recognized environmental features. The bottom line is that the climbing robots should always be able to keep moving on the wall without falling on the ground no matter how the environment map changes. This study proposes a contact force based motion planning scheme for transverse climbing robots to explore partially-known environments and adjust joint motion trajectories for persistent movements, just like a blind man using a white cane to detect obstacles in the walking path. The study adopts the technique of momentum observers to estimate the collision force between the robot’s exploring hand and the environment, in which a time varying threshold criterion is integrated into the contact detection algorithm to alleviate the effects of model uncertainties and noises. Once the robot detects a collision, the system then smoothly revises the robot’s locomotion with joint trajectories calculated by an admittance control law to overcome the environmental transition. For the case of robot climbing in a free space, i.e., the exploring hand finds no graspable targets from the previous move, an A* based searching algorithm will be applied to identify the next feasible moving path along with proper locomotion adjustment. Finally, the above methods are implemented on a two-link robot transversely climbing along the horizontal ledges but with arbitrary elevation or inclination changes. The simulation successfully justifies the effectiveness of the developed contact force based motion planning algorithm. The proposed idea of “touch and move” exploration strategies could be also applied to developing advanced motion planning methodologies for bio-inspired climbing robots such as brachiation robots.

[1] 貝爾網站國際有限公司, “案例實績 - 合發防水抓漏工程|大台北屋頂抓漏|外牆防水抓漏-防水漆.” https://www.hefas.com.tw/case_detail.php?id=96 (accessed Jun. 13, 2023).
[2] https://www.facebook.com/kknews.cc, “工業清洗——外牆清洗應注意的事項！.” https://kknews.cc/home/r85y29o.html (accessed Jun. 13, 2023).
[3] “Climbing Robot - an overview | ScienceDirect Topics.” https://www.sciencedirect.com/topics/engineering/climbing-robot (accessed Jul. 11, 2023).
[4] 台灣的8大住宅"特色",” 好房網News, Oct. 09, 2014. https://news.housefun.com.tw/news/article/30110381124.html (accessed Jun. 20, 2023).
[5] T. Fukuda, H. Hosokai, and Y. Kondo, “Brachiation type of mobile robot,” in Fifth International Conference on Advanced Robotics ’Robots in Unstructured Environments, Jun. 1991, pp. 915–920 vol.2. doi: 10.1109/ICAR.1991.240556.
[6] 羅威竣, “參考壁面凸桿攀爬者運動步態之橫向抓枝機器人設計與實現, ” 國立臺灣科技大學機械工程系, 2021
[7] 楊宗翰, “橫向飛躍抓枝機器人之設計改良與運動控制研究.” 國立臺灣科技大學機械工程系, 2018
[8] Z. Zheng et al., “CCRobot-IV: An Obstacle-Free Split-Type Quad-Ducted Propeller-Driven Bridge Stay Cable-Climbing Robot,” IEEE Robot. Autom. Lett., vol. 7, no. 4, pp. 11751–11758, Oct. 2022, doi: 10.1109/LRA.2021.3099872.
[9] C.-Y. Lin and Y.-J. Tian, “Design of Transverse Brachiation Robot and Motion Control System for Locomotion between Ledges at Different Elevations,” Sensors, vol. 22, no. 11, Art. no. 11, Jan. 2022, doi: 10.3390/s22114031.
[10] “猴子 爬树图片_猴子 爬树图片下载_正版高清图片库_第5页-Veer图库.” https://www.veer.com/search-image/1147361/?sort=newest&page=5 (accessed Jul. 08, 2023).
[11] 胡婷鈞, “可於壁面凸塊進行橫向攀爬之抓技機器人設計與運動控制策略研究.”, 國立臺灣科技大學機械工程系, 2018.
[12] “Types of Sensors in Robotics,” Wevolver, 1673575060037. https://www.wevolver.com/article/sensors-in-robotics-the-common-types (accessed Jul. 11, 2023).
[13] M. Zhong, Y. Ma, Z. Li, J. He, and Y. Liu, “Façade Protrusion Recognition and Operation-Effect Inspection Methods Based on Binocular Vision for Wall-Climbing Robots,” Appl. Sci., vol. 13, no. 9, Art. no. 9, Jan. 2023, doi: 10.3390/app13095721.
[14] H. Eto and H. H. Asada, “Development of a Wheeled Wall-Climbing Robot with a Shape-Adaptive Magnetic Adhesion Mechanism,” in 2020 IEEE International Conference on Robotics and Automation (ICRA), May 2020, pp. 9329–9335. doi: 10.1109/ICRA40945.2020.9196919.
[15] “PyTorch.” https://www.pytorch.org (accessed Jul. 11, 2023).
[16] J.-H. Kim, J.-C. Lee, and Y.-R. Choi, “PiROB: Vision-based pipe-climbing robot for spray-pipe inspection in nuclear plants,” Int. J. Adv. Robot. Syst., vol. 15, no. 6, p. 1729881418817974, Nov. 2018, doi: 10.1177/1729881418817974.
[17] “CPG-Based Control of Humanoid Robot Locomotion | SpringerLink.” https://link.springer.com/referenceworkentry/10.1007/978-94-007-7194-9_49-1 (accessed Jul. 11, 2023).
[18] R. S. da Veiga, A. S. de Oliveira, L. V. R. de Arruda, and F. N. Junior, “Localization and Navigation of a Climbing Robot Inside a LPG Spherical Tank Based on Dual-LIDAR Scanning of Weld Beads,” in Robot Operating System (ROS): The Complete Reference (Volume 1), A. Koubaa, Ed., in Studies in Computational Intelligence. Cham: Springer International Publishing, 2016, pp. 161–184. doi: 10.1007/978-3-319-26054-9_7.
[19] N. Ahmed, “Wheel Odometry Model for Differential Drive Robotics,” Medium, Mar. 05, 2023. https://medium.com/@nahmed3536/wheel-odometry-model-for-differential-drive-robotics-91b85a012299 (accessed Jul. 01, 2023).
[20] “Sensor Fusion - an overview | ScienceDirect Topics.” https://www.sciencedirect.com/topics/engineering/sensor-fusion (accessed Jul. 01, 2023).
[21] M. P. Austin, M. Y. Harper, J. M. Brown, E. G. Collins, and J. E. Clark, “Navigation for Legged Mobility: Dynamic Climbing,” IEEE Trans. Robot., vol. 36, no. 2, pp. 537–544, Apr. 2020, doi: 10.1109/TRO.2019.2958207.
[22] B. Saund, S. Choudhury, S. Srinivasa, and D. Berenson, “The blindfolded traveler’s problem: A search framework for motion planning with contact estimates,” Int. J. Robot. Res., p. 02783649231170893, May 2023, doi: 10.1177/02783649231170893.
[23] S. Zhong, N. Fazeli, and D. Berenson, “CHSEL: Producing Diverse Plausible Pose Estimates from Contact and Free Space Data.” arXiv, May 13, 2023. doi: 10.48550/arXiv.2305.08042.
[24] “Robotics: Useful definitions.” https://cs.stanford.edu/people/eroberts/courses/soco/projects/1998-99/robotics/definitions.html (accessed Jul. 08, 2023).
[25] 楊辰修, “應用機器學習辨識可供抓枝機器人抓握之凸起物場景.” 國立臺灣科技大學機械工程系, 2019
[26] Y. Li, Y. Li, M. Zhu, Z. Xu, and D. Mu, “A nonlinear momentum observer for sensorless robot collision detection under model uncertainties,” Mechatronics, vol. 78, p. 102603, Oct. 2021, doi: 10.1016/j.mechatronics.2021.102603.
[27] T. Fujiki and K. Tahara, “Series admittance–impedance controller for more robust and stable extension of force control,” ROBOMECH J., vol. 9, no. 1, p. 23, Dec. 2022, doi: 10.1186/s40648-022-00237-5.
[28] “Planning Algorithms / Motion Planning.” http://lavalle.pl/planning/ (accessed Jun. 25, 2023).
[29] “Path Planning - an overview | ScienceDirect Topics.” https://www.sciencedirect.com/topics/engineering/path-planning (accessed Jul. 01, 2023).
[30] “Introduction to A*.” http://theory.stanford.edu/~amitp/GameProgramming/AStarComparison.html
[31] “Lagrange Equation - an overview | ScienceDirect Topics.” https://www.sciencedirect.com/topics/engineering/lagrange-equation (accessed Jul. 11, 2023).
[32] T. Fujiki and K. Tahara, “Series admittance–impedance controller for more robust and stable extension of force control,” ROBOMECH J., vol. 9, no. 1, p. 23, Dec. 2022, doi: 10.1186/s40648-022-00237-5.
[33] J. H. Davis, “Luenberger Observers,” in Foundations of Deterministic and Stochastic Control, J. H. Davis, Ed., in Systems & Control: Foundations & Applications. Boston, MA: Birkhäuser, 2002, pp. 245–254. doi: 10.1007/978-1-4612-0071-0_8.
[34] “7.2: Generalized Momentum,” Physics LibreTexts, Nov. 11, 2017. https://phys.libretexts.org/Bookshelves/Classical_Mechanics/Variational_Principles_in_Classical_Mechanics_(Cline)/07%3A_Symmetries_Invariance_and_the_Hamiltonian/7.02%3A_Generalized_Momentum (accessed Jun. 15, 2023).
[35] P. J. From, J. T. Gravdahl, and K. Y. Pettersen, “Properties of the Dynamic Equations in Matrix Form,” in Vehicle-Manipulator Systems: Modeling for Simulation, Analysis, and Control, P. J. From, J. T. Gravdahl, and K. Y. Pettersen, Eds., in Advances in Industrial Control. London: Springer, 2014, pp. 285–305. doi: 10.1007/978-1-4471-5463-1_9.
[36] “Introduction to A*.” http://theory.stanford.edu/~amitp/GameProgramming/AStarComparison.html (accessed Jun. 23, 2023).
[37] “Dijkstra’s Shortest Path Algorithm - A Detailed and Visual Introduction,” freeCodeCamp.org, Sep. 28, 2020. https://www.freecodecamp.org/news/dijkstras-shortest-path-algorithm-visual-introduction/ (accessed Jun. 26, 2023).