液压并联机器人力/位混合控制策略研究

Hybrid Position/Force Control Scheme for Hydraulic Parallel Manipulator

针对液压驱动的六自由度并联机构,提出了基于机器视觉的力/位混合控制策略,以实现并联机构的跟随对接等需要与环境密切接触的任务。平台的视觉系统由一架CCD摄像机和一个激光测距器构成,用来测量平台与目标之间的相对距离,同时通过一个平板式六自由度力传感器对接触过程的作用力进行测量。其控制器由内外两个闭环构成,并通过补偿环节对外力扰动进行修正。最后通过若干试验验证整个控制方式的有效性。试验结果表明,整个系统在相对位置跟踪和接触力控制方面具有优异的性能。在整个对接试验过程中,接触力被控制在300 N之内。

The aim was to propose a combined position/force control scheme for Stewart parallel platform＇s interacting tasks with environment or human body based on force sensor and vision system.The vision system included a high resolution camera and a laser range finder which aimed at providing global position information. To enhance the performance of industrial tasks interacting with the environment,a novel hybrid position/force control algorithm using exteroceptive vision system for hydraulic Stewart parallel platform was proposed. A two control-loop scheme was designed for interacting tasks,and gravity compensation was demonstrated to balance the disturbance. A full experimental system was built and a number of experiments were performed. The whole experimental system consisted of two parts,which was the hydraulic parallel manipulator and the docking simulation system. Firstly,a relative position tracking experiment was conducted to investigate the manipulator＇s high-speed tracking ability.Movement commands of ± 100 mm/s in y axis and ± 80 mm/s in z axis were given to actuate the manipulator. From the experiment result,the relative position tracking errors of y and z axes were both smaller than 10 mm. And then the dynamic docking experiment was performed to investigate the interacting performance with the target object of the proposed combined position/force control law,where the movement speeds of the docking simulator was driven along the y axis at 50 mm/s and z axis at40 mm/s. In dynamic docking experiment,the contact force of the manipulator was controlled within300 N,the hydraulic manipulator could track and dock with the moving target reliably and quickly. The above two experiments were conducted and the experiment results demonstrated the effectiveness and stability of the proposed algorithm.