【作者】 魏秀权；

【导师】 吴林；

【作者基本信息】 哈尔滨工业大学 ， 材料加工工程， 2009， 博士

【摘要】 核环境设备修复、海洋工程水下施工、空间站建设维护以及地下管道维修等极限环境任务,需要操作者在远离工作现场的安全环境中,根据现场反馈的传感信息对焊接设备和焊接过程进行远程监视和控制,实现遥控焊接。对于极限环境管道裂纹修复任务,采用基于管道替换策略的机器人遥控焊接维修方案具有很好的应用前景。本文针对基于力觉传感控制的非结构化环境装配策略进行研究,用于实现管道替换遥控焊接方案中涉及的作业工具装配操作,为机器人遥控焊接技术在极限环境管道维修领域的实际应用奠定基础。根据管道替换遥控焊接方案中工具装配操作的过程特点,设计开发了基于开放式位置伺服机器人、六维力/力矩传感器和力反馈操纵杆的机器人遥控焊接力觉传感控制系统,作为表面跟踪共享力觉控制和工具装配主动柔顺控制等力觉传感控制关键技术的研究平台。力觉传感控制技术是机器人遥控焊接非结构化环境装配策略研究的核心问题。通过对遥控焊接任务空间和人机智能分配模式的分析,提出了被动柔顺共享监督和宏观遥控局部自主两类非结构化环境装配策略。被动柔顺共享监督装配策略采用表面跟踪共享力觉控制任务空间标定技术将非结构化环境转化为结构化环境,通过人机交互任务分析规划技术和人机共享遥操作辅助功能,实现工具装配任务的共享监督控制,以被动柔顺方式补偿装配误差。宏观遥控局部自主装配策略通过宏观遥控操作引导装配工具与任务环境发生接触,根据工具重力补偿后的实际接触力信息进行基于受控Petri网模型的工具装配过程离散事件控制,通过主动柔顺控制消除装配位置姿态偏差。对表面跟踪共享力觉控制任务空间标定技术进行了研究,采用基于球形力探测头和给定约束运动平面的任务环境表面跟踪策略,根据跟踪轨迹接触点受力状态分析,提出了位置伺服机器人表面跟踪局部自主力控制算法,建立了基于人机智能协作的表面跟踪共享力觉控制系统,用于实现任务空间标定。通过基于坐标系变换的接触点测量算法,精确计算跟踪轨迹接触点的空间位置。采用直接最小二乘拟合算法,对管道工件表面的椭圆跟踪轨迹进行拟合。提出了针对管道工件的任务空间标定算法,用于建立满足管道维修遥控焊接被动柔顺装配精度要求的结构化环境模型。对于被动柔顺共享监督装配策略,采用智能分级思想和基元任务概念,进行人机交互任务分析规划器设计,用于构造实现监督控制的基元任务序列。针对受限运动空间装配任务,提出了基于平面辅助功能和直线—速度—力辅助功能的人机共享控制策略,提高了任务分析规划器的适应能力。开发了基元任务编译器,将监督控制高级指令序列解释为结构体格式的中间指令序列,进而生成机器人指令序列,以被动柔顺方式完成工具装配任务。对基于负载参数标定的工具重力补偿技术进行了研究,在无接触只存在重力作用的条件下,通过工具重力计算和线性最小二乘进行工具负载参数标定。在此基础上,提出了用于消除主动柔顺装配过程重力干扰的工具重力补偿算法,使重力补偿精度满足工具装配任务局部自主力控制的需要。对于宏观遥控局部自主装配策略,在接触形式描述和接触状态分析的基础上,构造基于受控Petri网的管道工具装配接触状态离散事件模型,提出了基于状态标识检测的装配接触状态变迁识别方法,设计了用于实现主动柔顺工具装配的离散事件控制器,根据工具重力补偿后的实际接触力信息进行工具装配离散事件控制,以主动柔顺方式消除工具装配位置姿态偏差。针对基于管道替换策略的极限环境管道维修遥控焊接任务,分别采用被动柔顺共享监督和宏观遥控局部自主策略进行非结构化环境工具装配实验。从操作精度、执行效率、通用性和可靠性角度进行对比分析的结果表明,被动柔顺共享监督策略的通用性和可靠性较好,宏观遥控局部自主策略的操作精度和执行效率较高。针对机器人遥控焊接非结构化环境力觉辅助装配策略的研究工作为解决遥控焊接管道维修涉及的工具装配操作问题提供了切实可行的途径。更多还原

【Abstract】 For extreme environment tasks such as maintenance in nuclear power plants, marine engineering undersea, construction of space stations and pipelines repairing underground, it is necessary for operator to implement remote welding at a safe place away from the worksite, monitor the welding process and control the welding equipments with sensor feedback information from a distance. The maintenance scheme based on pipe segment replacement strategy with robotic remote welding illustrates a wide prospect of application for crack repairing task on pipelines in hazardous environment. In this dissertation, the assembly strategies for robotic remote welding in unstructured environment based on force sensing and control have been studied for the realization of tool assembly operation in remote welding scheme with pipe segment replacement, and the research work will be the foundation for futher practical application of robotic remote welding technology in extreme environment pipeline maintenance area.According to the features of tool assembly operation during pipeline maintenance process, a force sensing and control system for robotic remote welding is designed and developed with position-servo open architecture robot, six-dimensional force/torque sensor and force-feedback joystick, which is the research platform for key technologies of force sensing and control, such as shared force control for surface tracking and active compliant control for tool assembling. The core issue for research on robotic remote welding assembly strategy is how to use the key technologies to implement assembly task in unstructured environment.Based on the analysis of remote welding task space and human-machine intelligence distribution mode, the strategies of shared supervisory with passive compliance and local autonomous with global teleoperation are proposed for assembly task in unstructured environment. For shared supervisory strategy with passive compliance, task space calibration technique based on surface tracking with shared force control is adopted to turn the unstructured environment into a structured environment. The shared supervisory control of assembly task is achieved by human-computer interactive task analyzing and planning technology and human-machine shared teleoperation assistance function. The assembly error is compensated by passive compliant device. For local autonomous strategy with global teleoperation, the assembly tool is manipulated in global teleoperation mode to contact with the task environment. The discrete event control of tool assembly process is based on the actual contact force information with tool gravity compensation and contact state model with controlled Petri Net. The deviation of assembly position and orientation is eliminated by active compliant control.Task space calibration technique based on surface tracking with shared force control is studied, while the task environment surface tracking strategy based on spherical force probe and constrained motion plane is adopted. According to the force status analysis of contact point on tracking trajectory, a local autonomous force control algorithm for surface tracking with position-servo robot is proposed, and the shared force control system for surface tracking is established to realize task space calibration. Based on the algorithm of coordinate system transformation, the accurate spatial locations of contact points on tracking trajectory are calculated. A task space calibration algorithm for pipe workpiece, in which the elliptic tracking trajectory on pipe surface are fitted by direct least squares algorithm, is presented to make structured environment model meet the requirements of passive compliant assembly accuracy for pipe maintenance remote welding.Shared supervisory strategy with passive compliance for assembly task is based on the concept of hierarchical intelligence and atomic task. A human-computer interactive task analyzer and planner is designed and implemented to construct the supervisory control command sequence of atomic tasks. For assembly task in restricted motion space, a human-machine shared control strategy based on planar and linear-velocity-force teleoperation assistance function is proposed to improve the adaptability and reliability of task analyzing and planning system. An atomic task compiler is developed to interpret high-level command sequence into intermediate struct-format command sequence. Robot command sequence is generated to achieve tool assembly task with passive compliance.The load parameters calibration algorithm is based on tool gravity calculation and direct least squares solution under non-contact gravity condition. Tool gravity compensation technique is studied to eliminate the gravity disturbance during active compliant tool assembly process, and make the precision of gravity compensation meet the requirements of local autonomous force control for tool assembly task.Local autonomous strategy with global teleoperation for assembly task is based on contact formation description and contact state analysis. A discrete event controlled Petri-net model of tool assembly contact state is constructed. An identification method of assembly contact state transition with state mark checking is proposed. A discrete event controller is designed to realize local autonomous compliant control of tool assembly process which is based on the actual contact force information with tool gravity compensation. The deviation of assembly position and orientation during tool assembly operation is gradually eliminated by active compliant control.Based on pipe segment replacement strategy of crack repairing tasks on pipelines in extreme environment with remote welding, the experiments on assembly operation in unstructured environment are performed with shared supervisory strategy with passive compliance and local autonomous strategy with global teleoperation respectively. The comparative analysis is made with respect to the operation accuracy, execution efficiency, adaptability and reliability. The analysis results show that, the adaptability and reliability of shared supervisory strategy with passive compliance are greater, while the operation accuracy and execution efficiency of local autonomous strategy with global teleoperation is higher. The research work on force assisted assembly strategies for robotic remote welding in unstructured environment provide a practical solution of tool assembly operation during remote welding for pipeline maintenance.更多还原