基于区域姿态解算的五指机械手抓取系统设计与实现  

Design and Implementation of the Five-fingered Robotic Hand GraspingSystem Based on Regional Pose Solving

作　　者：禹鑫燚[1] 赵崇良 陈磊 朱明珠 欧林林[1] YU Xinyi;ZHAO Chongliang;CHEN Lei;ZHU Mingzhu;OU Linlin(College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023,China;School of Astronautics,Northwestern Polytechnical University,Xi’an 710072,China)

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310023 [2]西北工业大学航天学院,陕西西安710072

出　　处：《机器人》2023年第6期698-709,共12页Robot

基　　金：浙江省自然科学基金(LY21F030018,LQ22F030021);国家自然科学基金(52005440)。

摘　　要：针对五指机械手抓取成功率低和抓取任务相对简单的问题,基于区域姿态解算方法设计并实现了一种满足不同复杂任务的五指抓取系统。首先,设计了一种气动五指软爪,该软爪使用多种不同刚度的材料制成,由1根主气管和5根支气管驱动,控制复杂度较低,机械性能和抓取性能良好,适用不同的抓取策略。进一步结合软爪的特点提出了基于区域姿态解算的抓取策略。通过预测人手抓取物体时在物体上的接触区域,求解接触区域与软爪指尖在空间上的姿态解,计算软爪的抓取姿态和关节弯曲角度,该策略能够生成高鲁棒性的抓取姿态。然后,设计了包含大量物体的接触数据集。对数据集中的物体标注人手在抓取操作中指尖的接触区域,并尽可能地去除场景信息,提高了数据集的通用性,可用作基准数据集测试算法性能。最后,设计了一系列实验来验证软爪和抓取策略在复杂场景下的抓取性能,实验结果表明了所设计的五指软爪抓取系统在复杂场景下的有效性和可靠性。In order to address the low success rate and limited task capability of five-fingered manipulators,a five-fingered grasping system is designed and developed based on regional pose solving,to handle various complex grasping tasks.Firstly,a pneumatic five-fingered soft hand is designed,which is made from various materials with varying levels of stiffness,and is powered by one main and five branch tracheae,offering excellent mechanical and grasping performances,with reduced control complexity,and the ability to implement different grasping strategies.Further,a grasping strategy based on regional pose solving is proposed by considering the characteristics of soft grippers.This strategy generates robust grasping poses by predicting the contact area of the human hand when grasping an object,solving the spatial pose of the contact area and soft hand fingertips,and calculating the grasping pose and bending angle of the soft hand.Then,a contact dataset with a large number of objects is created,marking the contact areas of the human fingertips during grasping the objects in the dataset,and minimizing scene information for increased generality.This dataset can be used as a benchmark to evaluate the performance of different algorithms.Finally,a series of experiments are designed to verify the grasping performance of the five-fingered soft hand and the grasping strategy in complex scenarios.The experimental results demonstrate the effectiveness and reliability of the designed five-fingered soft hand grasping system in complex environments.

关 键 词：机器人抓取 五指软爪 姿态解算 抓取数据集 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]