Characterisation of Grasp Quality Metrics

Robot grasp quality metrics are used to evaluate, compare and select robotic grasp configurations. Many of them have been proposed based on a diversity of underlying principles and to assess different aspects of the grasp configurations. As a consequence, some of them provide similar information but other can provide completely different assessments. Combinations of metrics have been proposed in order to provide global indexes, but these attempts have shown the difficulties of merging metrics with different numerical ranges and even physical units. All these studies have raised the need of a deeper knowledge in order to determine independent grasp quality metrics which enable a global assessment of a grasp, and a way to combine them. This paper presents an exhaustive study in order to provide numerical evidence for these issues. Ten quality metrics are used to evaluate a set of grasps planned by a simulator for 7 different robot hands over a set of 126 object models. Three statistical analysis, namely, variability, correlation and sensitivity, are performed over this extensive database. Results and graphs presented allow to set practical thresholds for each quality metric, select independent metrics, and determine the robustness of each metric,providing a reliability indicator under pose uncertainty. The results from this paper are intended to serve as guidance for practical use of quality metrics by researchers on grasp planning algorithms.     

Development of a multifingered robotic hand with the thenar grasp function

ABSTRACT

Manipulation of objects after grasping is an important research topic for robotic hands. In this research, we propose a grip method based on the thenar opposition, which is rarely handled in the framework of conventional precision and power grasps and useful in withdrawing a large torque after the grip, and develop a robotic hand that can realize the grasp as well as normal grasps. To this end, we propose methods to evaluate the thenar opposition and the variety of grasps based on the joint alignment of the thumb. Based on these methods, we determine the kinematic structure of the robotic hand and develop it with gear train transmissions and a soft skin. Experiments are implemented to demonstrate that the robotic hand realizes various grasps and a gripping force sufficient to withstand the torque that opens tightly clamped hand valves.     

基于深度神经网络的多指灵巧手抓取手势优化

基于深度神经网络模型,提出了一种适用于多指灵巧手的抓取手势优化方法。首先,在仿真环境下构建了一个抓取数据集,并在此基础上训练了一个卷积神经网络,依据目标物体单目视觉信息和多指灵巧手抓取位形来预测抓取质量函数,由此可以将多指灵巧手的抓取规划问题转化为使抓取质量最大化的优化问题,进一步,基于深度神经网络中的反向传播和梯度上升算法实现多指灵巧手抓取手势的迭代与优化。在仿真环境中,比较该网络和仿真平台对同一抓取位形的抓取质量评估结果,再利用所提出的优化方法对随机搜索到的初始手势进行优化,比较优化前后手势的力封闭指标。最后,在实际机器人平台上验证本文方法的优化效果,结果表明,本文方法对未知物体的抓取成功率在80%以上,对于失败的抓取,优化后成功的比例达到90%。     

Recognition of human grasps by time-clustering and fuzzy modeling

In this paper, we address the problem of recognition of human grasps for five-fingered robotic hands and industrial robots in the context of programming-by-demonstration. The robot is instructed by a human operator wearing a data glove capturing the hand poses. For a number of human grasps, the corresponding fingertip trajectories are modeled in time and space by fuzzy clustering and Takagi–Sugeno (TS) modeling. This so-called time-clustering leads to grasp models using time as an input parameter and fingertip positions as outputs. For a sequence of grasps, the control system of the robot hand identifies the grasp segments, classifies the grasps and generates the sequence of grasps shown before. For this purpose, each grasp is correlated with a training sequence. By means of a hybrid fuzzy model, the demonstrated grasp sequence can be reconstructed.     

Learning how to grasp based on neural network retraining

Humans have an incredible capacity to manipulate objects using dextrous hands. A large number of studies indicate that robot learning by demonstration is a promising strategy to improve robotic manipulation and grasping performance. Concerning this subject we can ask: How does a robot learn how to grasp? This work presents a method that allows a robot to learn new grasps. The method is based on neural network retraining. With this approach we aim to enable a robot to learn new grasps through a supervisor. The proposed method can be applied for 2D and 3D cases. Extensive object databases were generated to evaluate the method performance in both 2D and 3D cases. A total of 8100 abstract shapes were generated for 2D cases and 11700 abstract shapes for 3D cases. Simulation results with a computational supervisor show that a robotic system can learn new grasps and improve its performance through the proposed HRH (Hopfield-RBF-Hopfield) grasp learning approach.     

OCOG: A common grasp computation algorithm for a set of planar objects

This paper addresses the problem of defining a simple End-Effector design for a robotic arm that is able to grasp a given set of planar objects. The OCOG (Objects COmmon Grasp search) algorithm proposed in this paper searches for a common grasp over the set of objects mapping all possible grasps for each object that satisfy force closure and quality criteria by taking into account the external wrenches (forces and torque) applied to the object. The mapped grasps are represented by feature vectors in a high-dimensional space. This feature vector describes the design of the gripper. A database is generated for all possible grasps for each object in the feature vector space. A search algorithm is then used for intersecting all possible grasps over all parts and finding a common grasp suitable for all objects. The search algorithm utilizes the kd-tree index structure for representing the database of the sets of feature vectors. The kd-tree structure enables an efficient and low cost nearest-neighbor search for common vectors between the sets. Each common vector found (feature vector) is the grasp configuration for a group of objects, which implies the future end-effector design. The final step classifies the grasps found to subsets of the objects, according to the common vectors found. Simulations and experiments are presented for four objects to validate the feasibility of the proposed algorithm. The algorithm will be useful for standardization of end-effector design and reducing its engineering time.     

Grasp programming by demonstration in virtual reality with automatic environment reconstruction

A virtual reality system enabling high-level programming of robot grasps is described. The system is designed to support programming by demonstration (PbD), an approach aimed at simplifying robot programming and empowering even unexperienced users with the ability to easily transfer knowledge to a robotic system. Programming robot grasps from human demonstrations requires an analysis phase, comprising learning and classification of human grasps, as well as a synthesis phase, where an appropriate human-demonstrated grasp is imitated and adapted to a specific robotic device and object to be grasped. The virtual reality system described in this paper supports both phases, thereby enabling end-to-end imitation-based programming of robot grasps. Moreover, as in the PbD approach robot environment interactions are no longer explicitly programmed, the system includes a method for automatic environment reconstruction that relieves the designer from manually editing the pose of the objects in the scene and enables intelligent manipulation. A workspace modeling technique based on monocular vision and computation of edge-face graphs is proposed. The modeling algorithm works in real time and supports registration of multiple views. Object recognition and workspace reconstruction features, along with grasp analysis and synthesis, have been tested in simulated tasks involving 3D user interaction and programming of assembly operations. Experiments reported in the paper assess the capabilities of the three main components of the system: the grasp recognizer, the vision-based environment modeling system, and the grasp synthesizer.     

Hierarchical reinforcement learning of multiple grasping strategies with human instructions

Grasping is an essential component for robotic manipulation and has been investigated for decades. Prior work on grasping often assumes that a sufficient amount of training data is available for learning and planning robotic grasps. However, constructing such an exhaustive training dataset is very challenging in practice, and it is desirable that a robotic system can autonomously learn and improves its grasping strategy. Although recent work has presented autonomous data collection through trial and error, such methods are often limited to a single grasp type, e.g. vertical pinch grasp. To address these issues, we present a hierarchical policy search approach for learning multiple grasping strategies. To leverage human knowledge, multiple grasping strategies are initialized with human demonstrations. In addition, a database of grasping motions and point clouds of objects is also autonomously built upon a set of grasps given by a user. The problem of selecting the grasp location and grasp policy is formulated as a bandit problem in our framework. We applied our reinforcement learning to grasping both rigid and deformable objects. The experimental results show that our framework autonomously learns and improves its performance through trial and error and can grasp previously unseen objects with a high accuracy.     

基于多模特征深度学习的机器人抓取判别方法

针对智能机器人抓取判别问题,研究多模特征深度学习与融合方法.该方法将测试特征分布偏离训练特征视为一类噪化,引入带稀疏约束的降噪自动编码（Denoising auto-encoding, DAE）,实现网络权值学习;并以叠层融合策略,获取初始多模特征的深层抽象表达,两种手段相结合旨在提高深度网络的鲁棒性和抓取判别精确性.实验采用深度摄像机与6自由度工业机器人组建测试平台,对不同类别目标进行在线对比实验.结果表明,设计的多模特征深度学习依据人的抓取习惯,实现最优抓取判别,并且机器人成功实施抓取定位,研究方法对新目标具备良好的抓取判别能力.     

Automated Gripper Jaw Design and Grasp Planning for Sets of 3D Objects

An algorithm for automatically generating a common jaw design and planning grasps for a given set of polyhedral objects is presented. The algorithm is suitable for a parallel‐jaw gripper equipped with three cylindrical fingers. The common jaw design eliminates the need for custom made grippers and tool changing. The proposed jaw configuration and planning approach reduces the search associated with locating the finger contacts from six degrees‐of‐freedom to one degree‐of‐freedom. Closed‐form algorithms for checking force closure and for predicting jamming are developed. Three quality metrics are introduced to improve the quality of the planned grasps. The first is a measure of the sensitivity of the grasp to errors between the actual and planned finger locations. The second is a measure of the efficiency of the grasp in terms of the contact forces. The third is a measure of the dependence of force closure on friction. These quality metrics are not restricted to cylindrical fingers and can be applied to n finger grasps. Running on a standard PC, the algorithm generated a solution in less than five minutes for a set of five objects with a total of 456 triangular facets. © 2003 Wiley Periodicals, Inc.     

Vision-based three-finger grasp synthesis constrained by hand geometry

This paper addresses the problem of designing a practical system able to grasp real objects with a three-fingered robot hand. A general approach for synthesizing two- and three-finger grasps on planar unknown objects is presented. Visual perception is used to reduce the uncertainty and to obtain relevant information about the objects.We focus on non-modeled planar extruded objects, which can represent many real-world objects. In addition, particular mechanical constraints of the robot hand are considered.First, a vision processing module that extracts from object images the relevant information for the grasp synthesis is developed. This is completed with a set of algorithms for synthesizing two- and three-finger grasps taking into account force-closure and contact stability conditions, with a low computational effort. Finally, a procedure for constraining these results to the kinematics of the particular hand, is also developed. In addition, a set of heuristic metrics for assessing the quality of the computed grasps is described.All these components are integrated in a complete system. Experimental results using the Barrett hand are shown and discussed.     

Interactive teaching of task-oriented robot grasps

This paper focuses on the problem of grasp stability and grasp quality analysis. An elegant way to evaluate the stability of a grasp is to model its wrench space. However, classical grasp quality measures suffer from several disadvantages, the main drawback being that they are not task related. Indeed, constructive approaches for approximating the wrench space including also task information have been rarely considered. This work presents an effective method for task-oriented grasp quality evaluation based on a novel grasp quality measure. We address the general case of multifingered grasps with point contacts with friction.The proposed approach is based on the concept of programming by demonstration and interactive teaching, wherein an expert user provides in a teaching phase a set of exemplar grasps appropriate for the task. Following this phase, a representation of task-related grasps is built. During task planning and execution, a grasp could be either submitted interactively for evaluation by a non-expert user or synthesized by an automatic planning system. Grasp quality is then assessed based on the proposed measure, which takes into account grasp stability along with its suitability for the task. To enable real-time evaluation of grasps, a fast algorithm for computing an approximation of the quality measure is also proposed. Finally, a local grasp optimization technique is described which can amend uncertainties arising in supplied grasps by non-expert users or assist in planning more valuable grasps in the neighborhood of candidate ones.The paper reports experiments performed in virtual reality with both an anthropomorphic virtual hand and a three-fingered robot hand. These experiments suggest the effectiveness and task relevance of the proposed grasp quality measure.     

机器人多指手灵巧抓持规划

抓持规划是机器人灵巧手要完成预期任务所面临的一个重要问题．本文采用主从操作方式进行灵巧手的指尖抓持规划，由人手决定抓持接触点的位置， 灵巧手通过调整其手掌的位置和姿态保证各手指在人手指定的位置上抓持物体．根据灵巧手的操作特点，提出以关节灵活度来描述关节运动各向同性的能力，并据此定义灵巧手操作灵活度，作为灵巧手抓持位形性能的评价指标．以最大操作灵活度作为优化目标函数，寻求最优的抓持性能．同时，借鉴人手的抓持经验，通过主从操作方式，建立从人手到灵巧手的运动映射关系，从而为手掌位置优化问题提供合理的初值．仿真实验结果说明了文中方法的有效性．     

On the role of hand synergies in the optimal choice of grasping forces

Recent work on the analysis of natural and robotic hands has introduced the notion of postural synergies as a principled organization of their complexity, based on the physical characteristics of the hand itself. Such characteristics include the mechanical arrangements of joints and fingers, their couplings, and the low-level control reflexes, that determine the specific way the concept of ??hand?? is embodied in a human being or a robot. While the focus of work done so far with postural synergies has been on motion planning for grasp acquisition, in this paper we set out to investigate the role that different embodiments have on the choice of grasping forces, and on the ultimate quality of the grasp. Numerical results are presented showing quantitatively the role played by different synergies (from the most fundamental to those of higher-order) in making a number of different grasps possible. The effect of number and types of engaged synergies on the distribution of optimal grasp forces is considered. Moreover, robustness of results is investigated with respect to variation in uncertain parameters such as contact and joint stiffness.     

Graspit! A versatile simulator for robotic grasping

A robotic grasping simulator, called Graspit!, is presented as versatile tool for the grasping community. The focus of the grasp analysis has been on force-closure grasps, which are useful for pick-and-place type tasks. This work discusses the different types of world elements and the general robot definition, and presented the robot library. The paper also describes the user interface of Graspit! and present the collision detection and contact determination system. The grasp analysis and visualization method were also presented that allow a user to evaluate a grasp and compute optimal grasping forces. A brief overview of the dynamic simulation system was provided.     

Data-driven grasp synthesis using shape matching and task-based pruning

Human grasps, especially whole-hand grasps, are difficult to animate because of the high number of degrees of freedom of the hand and the need for the hand to conform naturally to the object surface. Captured human motion data provides us with a rich source of examples of natural grasps. However, for each new object, we are faced with the problem of selecting the best grasp from the database and adapting it to that object. This paper presents a data-driven approach to grasp synthesis. We begin with a database of captured human grasps. To identify candidate grasps for a new object, we introduce a novel shape matching algorithm that matches hand shape to object shape by identifying collections of features having similar relative placements and surface normals. This step returns many grasp candidates, which are clustered and pruned by choosing the grasp best suited for the intended task. For pruning undesirable grasps, we develop an anatomically-based grasp quality measure specific to the human hand. Examples of grasp synthesis are shown for a variety of objects not present in the original database. This algorithm should be useful both as an animator tool for posing the hand and for automatic grasp synthesis in virtual environments.     

Efficient models for grasp planning with a multi-fingered hand

This paper presents a simple grasp planning method for a multi-fingered hand. Its purpose is to compute a context-independent and dense set or list of grasps, instead of just a small set of grasps regarded as optimal with respect to a given criterion. By context-independent, we mean that only the robot hand and the object to grasp are considered. The environment and the position of the robot base with respect to the object are considered in a further stage. Such a dense set can be computed offline and then used to let the robot quickly choose a grasp adapted to a specific situation. This can be useful for manipulation planning of pick-and-place tasks. Another application is human–robot interaction when the human and robot have to hand over objects to each other. If human and robot have to work together with a predefined set of objects, grasp lists can be employed to allow a fast interaction.The proposed method uses a dense sampling of the possible hand approaches based on a simple but efficient shape feature. As this leads to many finger inverse kinematics tests, hierarchical data structures are employed to reduce the computation times. The data structures allow a fast determination of the points where the fingers can realize a contact with the object surface. The grasps are ranked according to a grasp quality criterion so that the robot will first parse the list from best to worse quality grasps, until it finds a grasp that is valid for a particular situation.     

Classification of grasps by robot hands

Grasping robotic hands is classified into three categories based on the object connectivity. We decompose the space of contact forces into four subspaces and develop a method to determine the dimensions of the subspaces with respect to the connectivity of the grasped object. The relationships we obtain reveal the kinematic and static characteristics of three categories of grasps. It indicates how contact forces can be decomposed corresponding to each type of grasp. The technique also provides a guideline for determining the distribution of contact forces on grasped objects. We analyze how power grasps are identified from the object connectivity and used to synthesize hand configurations for grasping and manipulation tasks. A physical interpretation of the subspaces and the determination of their dimensions are illustrated by examples.     

Grasp synthesis for planar and solid objects

This article presents an analysis of the mechanics for multifingered grasps of planar and solid objects. We present a method that is intuitive and computationally efficient. We combine the search for finger grasp positions with finger (manipulation and squeezing) force calculations into a single method. Physically, the squeezing and frictional effects between the fingers and the grasped objects are fully visualized through our approach. Mathematically, we reduced the complexity of finger force calculations when comparing our scheme with previously available schemes. Examples are used to illustrate the effectiveness and efficiency of our scheme. Based upon the analysis of grasp mechanics, an algorithm for quantatively choosing the grasp points is proposed to ensure stable grasps.     

Compliant grasping with passive forces

Because friction is central to robotic grasp, developing an accurate and tractable model of contact compliance, particularly in the tangential direction, and predicting the passive force closure are crucial to robotic grasping and contact analysis. This paper analyzes the existence of the uncontrollable grasping forces (i.e., passive contact forces) in enveloping grasp or fixturing, and formulates a physical model of compliant enveloping grasp. First, we develop a locally elastic contact model to describe the nonlinear coupling between the contact force with friction and elastic deformation at the individual contact. Further, a set of “compatibility” equations is given so that the elastic deformations among all contacts in the grasping system result in a consistent set of displacements of the object. Then, combining the force equilibrium, the locally elastic contact model, and the “compatibility” conditions, we formulate the natural compliant model of the enveloping grasp system where the passive compliance in joints of fingers is considered, and investigate the stability of the compliant grasp system. The crux of judging passive force closure is to predict the passive contact forces in the grasping system, which is formulated into a nonlinear least square in this paper. Using the globally convergent Levenberg‐Marquardt method, we predict contact forces and estimate the passive force closure in the enveloping grasps. Finally, a numerical example is given to verify the proposed compliant enveloping grasp model and the prediction method of passive force closure. © 2005 Wiley Periodicals, Inc.