【作者】 丁亮；

【导师】 邓宗全；

【作者基本信息】 哈尔滨工业大学 ， 机械设计及理论， 2010， 博士

【摘要】 NASA的Sojourner、Spirit和Opportunity火星探测车取得了巨大的成就,拓展了人类对于火星的认识水平,同时在世界上掀起了利用轮式移动机器人(漫游车)进行星球探测的热潮。未来的星球探测任务(如MSL、ExoMars、“嫦娥”和SELENE)要求星球车能够自主运行于更加富有挑战性的松软崎岖地形环境当中。轮地相互作用地面力学可以广泛应用于星球车结构设计、性能评价、星壤参数辨识、动力学仿真、移动与导航控制等许多方面,是星球车性能提高的一个瓶颈问题,因而成为一个新的研究热点。在星球车研究过程中,目前主要直接应用传统车辆的地面力学成果。由于星球车与普通地面车辆在运行环境、控制方式、运行状态、载荷、车辆构型/尺寸和车轮构型/尺寸等方面都存在很大差异,而传统模型主要面向车辆设计且精度不高,因此针对星球探测车进行地面力学的实验、理论和应用方法研究是极其必要的。试验是地面力学研究的重要手段。在分析轮地相互作用力学影响因素的基础上进行试验设计,进而采用模拟月壤,利用高性能的车轮—土壤相互作用测试平台和El-dorado II四轮车测试系统进行系列试验,研究车轮尺寸(半径、宽度),轮刺尺寸(高度、个数、倾斜角度),地形信息(爬坡角度和侧偏角度),法向载荷,运行状态信息(滑转率,转向角、运行速度,重复通过次数)等因素对于轮地相互作用力学的影响,为后续理论分析、模型推导等提供基础研究数据。在深入分析车轮轮刺效应、滑转沉陷机理的基础上,建立高保真度的车轮—土壤相互作用滑转前进模型,并对载荷效应提出修正方法,利用实验数据验证了模型的有效性,并与传统模型进行了对比。基于上述成果,进一步推导了车轮滑移前进模型、侧偏模型、转向模型和前进与转向耦合模型,将崎岖地形中的车轮运动分解为爬坡和沿斜坡行走两种基本运动并进行了受力分析。对星球车轮地相互作用前进模型进行简化,构建了两种封闭解析解耦模型,提出了三种模型参数辨识方法,可以对反映星壤承压特性、剪切特性和轮地作用接触角的8个参数进行辨识,并进行试验验证,实现了对星壤特性的全面表征。基于积分模型的辨识方法具有较高的保真度,基于封闭解析解耦模型的辨识方法适用于参数的实时辨识。进一步推导了反映载荷效应的简化模型,利用El-Dorado II探测车的试验数据进行了Toyoura沙土的参数辨识。对车轮驱动性能评价的绝对及相对性能指标进行总结,包括沉陷指标、牵引能力指标和驱动电机性能指标,并推导公式揭示性能指标之间的内在联系。基于实验数据,从宏观和微观角度分析车轮宽度、半径,轮刺高度、个数和倾斜角度等设计参数对于其性能的影响,进而提出了车轮尺寸与轮刺的设计准则和方法。结合中国“嫦娥”探月工程中的月球车设计要求,进行了车轮设计和性能分析。以车体空间位姿和各关节角度作为广义坐标,采用递归方法建立运动学模型,推导了计算轮地接触点速度和构件质心速度的雅可比矩阵。利用Lagrange动力学方程、Newton-Euler方程等建立了融合轮地相互作用力学的通用星球车动力学模型。解决了车轮土壤相互作用接触区域和接触坐标系的求解这一关键难题。基于Matlab和SpaceDyn进行仿真程序实现,并利用整车测试数据进行验证。对松软崎岖地形中星球车的路径跟踪策略进行研究,建立了非完整运动学模型,设计了基于滑移补偿的转向控制策略进行路径跟踪。提出了基于轮刺痕迹和地面力学的两种车轮滑转率在线估计方法。分析了车轮滑转率与能量消耗的关系,证实了“等滑转率”是能量最优控制可以采用的次优解,利用车体速度前馈和反馈控制策略补偿滑转带来的速度损失。将上述算法融合,提出了时间—能量最优的星球车路径跟踪控制策略。利用仿真验证了算法的有效性和鲁棒性。本文推导了高保真度星球车轮地相互作用积分模型及封闭解析解耦模型,并成功应用于星壤参数辨识、车轮设计、高保真度动力学仿真和高性能移动控制,为松软崎岖地形环境中基于力学的星球车等移动机器人研究提供了解决方案。更多还原

【Abstract】 NASA’s Mars exploration rovers Sojourner, Sprit and Opportunity have achieved fruitful results and greatly widened the knowledge horizon of humankind, as a result of which an upsurge of exploring planets with wheeled mobile robots (rovers) was set up in the world. The future planet exploration missions, such as the MSL, ExoMars, Chang’e SELENE, require the rovers to traverse over more challenging deformable rough terrain than had ever encountered with limited supervision from the operator.Wheel-soil interaction terramechanics, which can be widely applied to planetary rover’s mechanical design, performance evaluation, soil parameter identification, dynamics simulation, mobility/navigation control, etc, is a bottle-neck problem for improving the performance of planetary rovers and becomes a hot research topic at the budding stage. During the research and development process of a planetary rover, the terramechanics knowledge for conventional terrestrial vehicles is usually used directly. However, there are many differences between the planetary rover and terrestrial vehicles from the aspects of running environment, control mode, running state, payload, chassis configuration/dimension and wheel type/dimension, etc. Moreover, the conventional models are oriented to vehicle design with relatively low precision. It is quite necessary to research on the terramechanics aiming at planetary rovers, including experiments, theory and application methods.Terramechanics is a subject that combines theoretical and experimental study closely. The factors that influence wheel-soil interaction terramechanics are analyzed in the beginning, according to which the experiments are designed. Then a wheel-soil interaction testbed and the El-Dorado II four-wheeled rover testbed are used for experimental study with lunar soil simulant. The influence on terramechanics caused by wheel dimensions (radius and width), lug parameters (height, number and inclination angle), terrain information (slope climbing angle and cross angle), normal load, running state information (slip ratio, steering angle, velocity, repetitive passing times) are tested, in order to provide basic data for further theoretical analysis and modeling.After analyzing the wheel lug effect and slip-sinkage principle, a high-fidelity driving model for a wheel moves forward with slip is derived, and a method for amending load effect is brought forward. The model is verified with experimental data. Based on it, the skid model for wheels moving forward, side slip model, steering model, and coupled model of moving forward and steering are deduced. The motion of a wheel moving on rough terrain is decomposed into two basic motions: climbing up/down and traversing across slopes and the mechanics is analyzed.The driving model of wheel-soil interaction terramechanics for a planetary rover’s wheel is simplified. Two kinds of closed-form analytical decoupled models are derived and three kinds of parameter identification methods are brought forward. Eight parameters that can reflect the bearing performance, shearing performance and contact angles could be identified to characterize the planetary soil comprehensively. They are verified with experimental data. The method that identifies soil parameters based on the integrated model has high fidelity. Methods that are developed based on closed-form analytical decoupled models, are suitable for real-time parameter identification. Simplified model considering load effect is deduced, based on which the parameters of Toyoura soil are identified with the experimental data obtained by El-Dorado II rover.Both the absolute and relative indices on evaluating a wheel’s driving performance are summarized, including sinkage indices, drawbar pull performance indices and motor performance indices, and equations are deduced to investigate the relationships among them. The influences of wheel radius/width and lug height/number/inclination angle on wheel performance are analyzed according to experimental data from micro and macro aspects, based on which the principles and methods for designing the dimensions and lugs of a wheel are brought forward. According to the mission requirements of China’s Chang’e lunar exploration project, lunar rover’s wheels are designed and the performances of them are analyzed.Kinematics equations are developed with recursive method, by using the position and orientation of rover’s body and the joint angles as generalized coordinates, and Jacobian matrices for calculating the velocities of wheel-soil interaction point and mass centers of all the components are deduced. Lagrange dynamics equation and Newton-Euler equation are then used to deduce generalized dynamics model combined with wheel-soil interaction mechanics. A key issue of calculating wheel-soil interaction area and coordinate is solved. The simulation is implemented with Matlab and SpaceDyn Toolbox, and verified with experimental data of El-Dorado II rover.Path following strategy is researched to control a rover moving in deformable rough terrain. Non-holonomic kinematics model is established and steering algorithm considering slip-compensation is designed. Two on-line slip ratio estimation methods are developed, based on lug traces and terramechanics, respectively. The relationship between energy consumption and slip ratio is analyzed. It is proved that“equal slip ratio”is a sub-optimal solution for energy optimal control. The velocity loss caused by wheel slip is compensated with feed-forward and feedback of rover body’s velocity. The control algorithms are combined and an energy-time optimal path following strategy for planetary rovers is brought forward. The control algorithms are verified by dynamics simulation using parameters of El-Dorado II rover and Toyoura sand.Wheel-soil interaction terramechanics models and closed-form analytical decoupled models with high-fidelity for planetary rover’s wheels are deduced, which are then successfully used for soil parameter identification, wheel design, high-fidelity dynamics simulation and high-performance locomotion control. The results of this study could provide solutions for the mechanics-based research of mobile robot, especially planetary rovers moving in deformable rough terrain.更多还原