王学林, 肖永飞, 毕淑慧, 范新建, 饶洪辉. 机器人柔性抓取试验平台的设计与抓持力跟踪阻抗控制[J]. 农业工程学报, 2015, 31(1): 58-63. DOI: doi:10.3969/j.issn.1002-6819.2015.01.009
    引用本文: 王学林, 肖永飞, 毕淑慧, 范新建, 饶洪辉. 机器人柔性抓取试验平台的设计与抓持力跟踪阻抗控制[J]. 农业工程学报, 2015, 31(1): 58-63. DOI: doi:10.3969/j.issn.1002-6819.2015.01.009
    Wang Xuelin, Xiao Yongfei, Bi Shuhui, Fan Xinjian, Rao Honghui. Design of test platform for robot flexible grasping and grasping force tracking impedance control[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2015, 31(1): 58-63. DOI: doi:10.3969/j.issn.1002-6819.2015.01.009
    Citation: Wang Xuelin, Xiao Yongfei, Bi Shuhui, Fan Xinjian, Rao Honghui. Design of test platform for robot flexible grasping and grasping force tracking impedance control[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2015, 31(1): 58-63. DOI: doi:10.3969/j.issn.1002-6819.2015.01.009

    机器人柔性抓取试验平台的设计与抓持力跟踪阻抗控制

    Design of test platform for robot flexible grasping and grasping force tracking impedance control

    • 摘要: 为减小机器人在采摘过程中对果蔬的损伤,设计了机器人柔性抓取试验平台,提出了一种适合末端执行器双指抓取果蔬的抓持力跟踪阻抗控制算法,该算法将末端执行器抓持果蔬系统等效为阻抗-导纳模型,使手指力/位控制等效为期望的惯量-阻尼-刚度模型,可按需调节其参数实现抓持力与位置的动态关系。期望抓持力与采集果蔬实际接触力的偏差作为外环力阻抗控制器的输入,控制器生成对内部位置环参考轨迹的校正量。该算法仅考虑沿末端执行器双指夹持果蔬方向,避免了使用多自由度机械臂阻抗控制算法的复杂性,提高了控制的实时性,同时对抓取系统模型的不确定和力扰动具有较强的鲁棒性。机器人抓取试验证明了双指抓持力反馈阻抗柔顺控制算法的有效性,可实现机器人柔性抓取,减小抓取果蔬损伤和保证品质。该研究可为农业机器人无损抓取和采摘提供关键技术。

       

      Abstract: Abstract: One of the major challenges of agricultural robots is the grasping and picking fresh fruit and vegetable without any damage under the complex environment. In order to minimize the harm due to robot grasp, this paper mainly introduces a dexterous multisensory gripper design and also develops a new force impedance control algorithm. First, the gripper with dual motor drive was designed, and each of fingers had an independent servo drive system and was integrated with force and tactile sensors. The calibration force sensors were mounted at the root of each finger, and the force signals were obtained by the force sensors. The expected force, position and speed parameters of the fingers can be set separately, so the grip centre and stroke can be controlled for the finger by pre-programming, the movement and forces of grasped objects can be actively controlled to achieve through the two-finger operations. The gripper with both the mechanical mechanism and control flexibility was mounted in the end of the industrial robot. The grasping experiment platform composed of industrial robot and dexterous gripper integrated monocular vision, force sensing and many kinds of sensors. The 2-D vision was able to quickly detect and locate grasped objects at the top of the experiment platform. Second, a force impedance control algorithm was proposed and used for one of fingers, position control was used for another finger, and it can regulate the grasping force by de?ning the target impedance between desired position and contact force. The whole grasping system can be equivalent to impedance & admittance model, and the finger force/position control can be equivalent to the expected target inertia-damping-stiffness model, and the model parameters can be adjustable according to the needs to realize the dynamic relationship between grasping force and position. The contact force errors between expected force value and actual force acquired from the force sensor were as the input of impedance controller and its output can be realized by the reference trajectory correction to the internal position control loop. The proposed algorithm only considered the direction of gripping fruit and vegetable, and the reference trajectory can be determined simply, and then avoids the use of complicated impedance control for multi-degree of freedom manipulators, through which it can improve the real-time control and robustness of the grasping system with model uncertainty or external force disturbance. Robot grasping experiments show that the system runs smoothly and reliably, the force-feedback impedance control is very effective, and the steady state error is maximum range within ± 0.4N in the experiment of grasping the tomatoes and eggs. It can make force track value with small force overshot and fast response simultaneously between the end-effector and fruit and vegetable, so it makes the force controller adaptive to the dynamic grasp process between the end-effector and fruit and vegetable, it can realize the flexible grasping and reduce damages and ensure the quality of the grasped fruit and vegetable. The research provides a key control technology for the compliant grasping of fruit and vegetable.

       

    /

    返回文章
    返回