Modeling, parameter identification and control of industrial tractor-trailers vehicles

A tractor towing multiple passive full trailers has been widely applied in industrial environments such as cargo terminals, warehouses and airports because of its higher effciency and lower cost compared with other cargo transportation vehicles. With the rapid development of autonomous driving technologies, a selfdriving tractor-trailers vehicle has attracted increasing attention and is promising to further improve the effciency and relieve the labor shortage issues. However, the autonomous control of the full-size tractor-trailers vehicle remains a challenge. The main diffculty is originated from the sophisticated system structure and the complex dynamics introduced by the heavy vehicle units and the payloads.The goal of this thesis is to achieve autonomous control of the tractor-trailers vehicle in real industrial environments with high accuracy and robustness. We first present the dynamic modeling of the tractor-trailers vehicle. To eliminate the high uncertainty of the kinematic and dynamic parameters of the trailers, we propose to install a force sensor to monitor the forces exerted on the tractor in real time. With the force measurements, the tractor can be considered separately and the dynamic model is significantly simplified. The inaccurate parameters and varying configurations of the trailers can all be ignored, while their dynamic effects on the tractor are still precisely captured. We then propose a Lyapunovbased trajectory tracking controller to online compensate the forces and drive the tractor onto the desired trajectory. The controller is developed using backstepping techniques and the stability is rigorously proved. An accurate vehicle mathematical model is essential for the development of self-driving algorithms and the precise parameters are also a prerequisite to implement our motion controller. To move towards the practical implementation, we further propose a comprehensive approach to identify key kinematic and dynamic parameters of a general car-like vehicle. The nonlinear vehicle-dynamics constraints are innovatively introduced into a factor-graph estimation framework that also fuses IMU and LiDAR measurements. The batch maximum a posterior (MAP) problem is formulated and solved effciently for both states and parameters. In addition, to generate more realistic reference motion instead of standard test trajectories, we propose a novel method to determine the configuration of the tractor-trailers vehicle in real time and to do trajectory planning in the online fashion. Finally, extensive experiments on our full-scale industrial tractor-trailers vehicle are conducted to verify the proposed methods. These field experiments in the warehouse and airport show satisfactory performance of our methods in the real industrial environments, which indicates a significant step towards the fully autonomous driving of the industrial tractor-trailers vehicles.拖曳多個被動式拖斗的拖車系統與其他貨運車輛相比具有更高的效率和更低的成本。因此這種拖車拖斗車輛被廣泛應用于諸如貨運站，倉庫和機場等工業環境中。隨著自動駕駛技術的飛速發展，無人駕駛的拖車拖斗系統已經引起越來越多的關注，並有望進一步提高效率並緩解勞動力短缺的問題。然而，對於這種拖車加拖斗車輛系統的自主控制仍然是一個很大的挑戰。主要困難源於車輛複雜的系統結構以及重型車輛單元和有效載荷帶來的複雜動力學特性。本文的目的是在工業環境中實現針對拖車拖斗車輛系統的高精度和高魯棒性的自主控制。本文首先提出了拖車拖斗系統的動力學建模。為了消除拖斗的運動學和動力學參數的高度不確定性，本文提出安裝一個力感測器以即時監控拖斗施加在拖車上的力，這樣我們就可以在動力學建模時單獨並且僅僅只考慮最前面的拖車，這可以大大簡化整個車輛系統的動力學模型。拖斗不準確的參數和變化的構型在控制的時候也都可以忽略不計，而同時它們對拖車的動態影響仍然可以被精確地捕捉到。根據動力學模型，本文隨後提出了一種基於李雅普諾夫的軌跡跟蹤控制器，該控制器可以實時補償後面拖斗作用在拖車上的力，並同時將拖車驅動到參考軌跡上。該控制器是採用反步法設計推導的，本文也嚴格證明了該控制器的穩定性。準確的車輛數學模型對於開發自動駕駛算法至關重要，精確的車輛參數也是實現本文運動控制器的前提。本文進一步提出了一種實用並且全面的方法來估計類汽車移動載體的關鍵運動學和動力學參數。本文創新性地將非線性車輛動力學約束引入到因子圖估計框架中，該框架還融合了慣性測量單元和三維激光雷達的測量。此外，為了在現實工業環境中生成更實際的參考運動而不是標準化的測試軌跡，我們提出了一種新穎的方法來實時估計拖車拖斗系統的構型並以線上方式進行軌跡規劃。通過線控改裝，我們開發了能夠自動駕駛的全尺寸工業無人拖車拖斗系統。基於改裝後的無人拖車平臺，我們針對本文所涉及的所有算法進行了大量的測試和實驗，驗證了所提出方法的有效性和優勢。這些在倉庫和機場進行的現場實驗顯示出我們的方法在實際工業環境中令人滿意的性能。這表明我們的研究是拖車拖斗系統邁向完全自主駕駛過程中的重要一步。Zhao, Hongchao.Ph.D. Chinese University of Hong Kong 2021.Includes bibliographical references (leaves ).Abstracts also in Chinese.Title from PDF title page (viewed on …)