Grasping Strategy for Unknown Objects Based on Real-Time Grasp-Stability Evaluation Using Proximity Sensing
Grasping Strategy for Unknown Objects Based on Real-Time Grasp-Stability Evaluation Using Proximity Sensing
复制标题
基于接近传感实时抓取稳定性评估的未知物体抓取策略
DOI：
10.1109/lra.2022.3188885
复制
发表时间：
2022
影响因子：
5.2
通讯作者：
Watanabe Tetsuyou
中科院分区：
文献类型：
--
作者：
Suzuki Yosuke;Yoshida Ryoya;Tsuji Tokuo;Nishimura Toshihiro;Watanabe Tetsuyou
这封信提出了一个强大的抓取策略，通过执行非接触抓取稳定性评估的基础上接近传感的未知物体。首先，我们开发了一个多指手的接近传感器。它的特点是传感器安装在一个球形指尖和透明树脂涂层，这有利于最近点估计和指尖运动生成。其次，我们实现了连续的评价抓持稳定性的基础上的力封闭假设最近的点是接触点。通过引入一种高速的传感和计算方法，可以连续评估和更新的抓持稳定性的手周围的物体移动。然后，我们实现了一个抓取策略，自适应激活多个运动基元的信息的基础上最近的点和抓取稳定性。它通过搜索具有高抓取稳定性的姿态并接近指尖以实现姿态来执行未知对象抓取。通过实际实验验证了该方法的有效性。
This letter proposes a robust grasping strategy for unknown objects by performing non-contact grasp-stability evaluation based on proximity sensing. First, we developed a proximity sensor for a multi-fingered hand. Its features are sensor mounting on a spherical fingertip and coating with transparent resin, which are advantageous for nearest point estimation and fingertip motion generation. Next, we realized consecutive evaluation of grasp-stability based on force-closure with assuming the nearest points to be the contact points. By introducing a high-speed sensing and calculation method, the grasp-stability can be consecutively evaluated and updated as the hand moves around the object. Then, we implemented a grasping strategy that adaptively activates multiple motion-primitives based on the information on the nearest points and the grasp-stability. It executes unknown-object grasping by searching for a pose with high grasp-stability and approaching the fingertip to realize the pose. The effectiveness of the methods was confirmed by actual experiments.