盲环境中基于触觉信息的灵巧手实时抓取策略的研究

项目名称： 盲环境中基于触觉信息的灵巧手实时抓取策略的研究

项目编号： No.61203346

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 自动化学科

项目作者： 樊绍巍

作者单位： 哈尔滨工业大学

项目金额： 24万元

中文摘要： 本项目以模拟人类在盲信息环境中抓取物体的操作方式为背景，探求灵巧手自主重构抓取对象模型、自动规划稳定抓取区域、自适应调整抓取控制力的方法。（1）基于指尖触觉单元接触态和非接触态信息特点，利用特征包（bag-of-fearture）的方法构建触觉单元接触位置/力信息的抓取特征识别库（2）基于手指触觉单元接触区域特征，提出基于专家系统的模糊神经网络混合智能模型，实现物体的模型重构及力封闭评价指标反馈器的动态修正。（3）基于指尖离散触觉单元的多链路精确运动学及动力学模型提出融合力封闭评价指标反馈器的抓取自适应阻抗控制策略，实现灵巧手的柔性触碰及稳定抓取。（4）构建盲信息环境下的自主抓取操作系统，完成对未知物体的模型重构及抓取操作实验。基于触觉多点信息的模型重构及抓取规划同控制系统集成的抓取操作方式不仅对多指协调操作领域具有重要的理论研究价值，同时能够有效提升灵巧手在非对称环境下抓取操作的能力。

中文关键词： 灵巧手；盲环境；触觉探索；模型重构；阻抗控制

英文摘要： The auto-recognition grasping model, the stable grasping region auto-planning and the grasping force auto-adjusting methods by the dexterous hand in the blind information environment are researched In this project. (1) The knowledge database between the tactile unit touch position and the force information using the bag-of-features method is constructed based on the advantages of the contact and noncontact model of the fingertip tactile cell. (2)A fuzzy-neural intelligent hybrid model based on expert system is put forward grounded on the local geometrical characteristic between the multi-fingertip tactile units and multi-touch points, in order to realize the model recognition of grasping objects and the dynamic correction of the force-closure evaluation index feedback. (3)The real-time multi-finger coordination adaptive resistance control strategy fused the force-closure evaluation index feedback is provided based on the multilink precise kinematic and dynamic models of the fingertip discrete touch points, in order to realize the flexible touch to the objects of the dexterous hand in the blind information environment and stable grasping during the manipulation period. (4) An auto-grasping manipulation system is founded in the blind environment, and recognized difference kind object, and the model recognition fo

英文关键词： Dexterous hands；blind environment；tactile exploration；model reconstruction；impedance control