Image Preprocessing-based Generalization and Transfer of Learning for Grasping in Cluttered Environments

International Journal of Control, Automation and Systems - In a cluttered environment in which objects are lying very closely to each other, the arranging motion is required before the robot...     

Grasping objects localized from uncertain point cloud data

Robotic grasping is very sensitive to how accurate is the pose estimation of the object to grasp. Even a small error in the estimated pose may cause the planned grasp to fail. Several methods for robust grasp planning exploit the object geometry or tactile sensor feedback. However, object pose range estimation introduces specific uncertainties that can also be exploited to choose more robust grasps. We present a grasp planning method that explicitly considers the uncertainties on the visually-estimated object pose. We assume a known shape (e.g. primitive shape or triangle mesh), observed as a–possibly sparse–point cloud. The measured points are usually not uniformly distributed over the surface as the object is seen from a particular viewpoint; additionally this non-uniformity can be the result of heterogeneous textures over the object surface, when using stereo-vision algorithms based on robust feature-point matching. Consequently the pose estimation may be more accurate in some directions and contain unavoidable ambiguities.The proposed grasp planner is based on a particle filter to estimate the object probability distribution as a discrete set. We show that, for grasping, some ambiguities are less unfavorable so the distribution can be used to select robust grasps. Some experiments are presented with the humanoid robot iCub and its stereo cameras.     

基于Camshift反馈码本模型的手抓取物跟踪方法

针对Camshift算法在复杂背景下无法自动跟踪目标的问题,提出一种基于Camshift反馈码本模型的运动目标检测和跟踪算法。该算法首先利用码本模型检测前景目标,然后采用Camshift在颜色概率分布图中跟踪前景区域中的目标,通过窗口尺寸比较和直方图相关性判断来解决自动跟踪,通过窗口位置预测和尺寸扩大来改进下一帧Camshift算法的输入搜索窗口,同时并集操作多个目标处理后的矩形窗口,并将其反馈为下一帧码本模型的图像检测区域。最后将该算法应用于手和目标物的抓取状态判断上,具体过程是在静态背景下利用两个摄像头采集到的图像进行手和目标物的检测和跟踪,通过矩形相交性判断抓取次数,以验证跟踪算法的有效性。实验结果表明,通过信息反馈减小了目标检测和跟踪的搜索区域,提高了算法的实时性,在单摄像头下可提高处理帧频130%。     

传递函数与脉冲传递函数的直接换算方法

在极点互异的条件下,本文导出了传递函数与脉冲传递函数参数换算的解析表达式,对不同的保持器,它们具有统一的数学形式和算法,对于重极点,本文给出一种有效的摄动方法,使其可用互异极点代替,作为应用例子,本文证明了应用Tustin算子的条件。     

Linear Momentum,Angular Momentum and Analytical Mechanics

The laws of mechanics can be set forth in different equivalent formulations. Frequently, one formulation can be deduced from another. For example, Lagrange's equations can be deduced both from Newton's law and from Hamilton's principle. One task never undertaken to date is that of deducing the principles of linear and angular momentum from a formulation of analytical mechanics. This task is undertaken in the present note in connection with simple nonholonomic systems.     

Momentum transfer to nanoobjects between isothermal parallel plates

A small-scale, trapezoidal rigid body in the gas-filled gap between two parallel plates at different temperatures is considered. An analytical expression for the thermally induced force onto the body in the direction parallel to the plates valid for an infinite Knudsen number is derived. For this purpose, diffuse reflection of the gas molecules at the solid walls is assumed. Simultaneously, Monte Carlo simulations are performed allowing an extension of the analysis to Knudsen numbers of the order of one. The numerical and the analytical results show excellent agreement, indicating that a temperature gradient orthogonal to the plates can induce a significant force in parallel direction, a phenomenon without analogy in the macroworld. This force is only slightly reduced when a Knudsen number of one is considered. In addition to the diffuse-reflection boundary condition, a mixture of diffuse and specular reflection is studied. The practical relevance of the results is exemplified by considering two scenarios with bodies of a specific geometry, among others a nanoscopic platelet.     

Perception and Grasping of Object Parts from Active Robot Exploration

Like humans, robots that need semantic perception and accurate estimation of the environment can increase their knowledge through active interaction with objects. This paper proposes a novel method for 3D object modeling for a robot manipulator with an eye-in-hand laser range sensor. Since the robot can only perceive the environment from a limited viewpoint, it actively manipulates a target object and generates a complete model by accumulation and registration of partial views. Three registration algorithms are investigated and compared in experiments performed in cluttered environments with complex rigid objects made of multiple parts. A data structure based on proximity graph, that encodes neighborhood relations in range scans, is also introduced to perform efficient range queries. The proposed method for 3D object modeling is applied to perform task-level manipulation. Indeed, once a complete model is available the object is segmented into its constituent parts and categorized. Object sub-parts that are relevant for the task and that afford a grasping action are identified and selected as candidate regions for grasp planning.     

Grasping the Torch

New Editor in Chief Roy Want, "grasping the torch" Mahadev Satyanarayanan handed to him, outlines the magazine's mission, offers a taste of things to come, and reveals his resolutions.     

Grasping an augmented object to analyse manipulative force control

Augmented reality allows changes to be made to the visual perception of object size even while the tangible components remain completely unaltered. It was, therefore, utilized in a study whose results are being reported here to provide the proper environment required to thoroughly observe the exact effect that visual change to object size had on programming fingertip forces when objects were lifted with a precision grip. Twenty-one participants performed repeated lifts of an identical grip apparatus to a height of 20 mm, maintained each lift for 8 seconds, and then replaced the grip apparatus on the table. While all other factors of the grip apparatus remained unchanged, visual appearance was altered graphically in a 3-D augmented environment. The grip apparatus measured grip and load forces independently. Grip and load forces demonstrated significant rates of increase as well as peak forces as the size of graphical images increased; an aspect that occurred in spite of the fact that extraneous haptic information remained constant throughout the trials. By indicating a human tendency to rely - even unconsciously - on visual input to program the forces in the initial lifting phase, this finding provides further confirmation of previous research findings obtained in the physical environment; including the possibility of extraneous haptic effects (Gordon et al. 1991a, Mon-Williams and Murray 2000, Kawai et al. 2000). The present results also suggest that existing knowledge concerning human manipulation tasks in the physical world may be applied to an augmented environment where the physical objects are enhanced by computer generated visual components.     

A Neural Network Approach to the Frictionless Grasping Problem

This article presents a heuristic technique used for solving linear complementarity problems(LCP). Determination of minimum forces needed to firmly grasp an object by a multifingered robot gripper for different external force and finger positions is our proposed application. The contact type is assumed to be frictionless. The interaction in the gripper–object system is formulated as an LCP. A numerical algorithm (Lemke) is used to solve the problem [3]. Lemke is a direct deterministic method that finds exact solutions under some constraints. Our proposed neural network technique finds almost exact solutions in solvable positions, and very good solutions for positions that Lemke fails to find solutions. A new adaptive technique is used for training the neural network and it is compared with the standard technique. Mathematical analysis for the convergence of the proposed technique is presented.     

Grasping an augmented object to analyse manipulative force control

Augmented reality allows changes to be made to the visual perception of object size even while the tangible components remain completely unaltered. It was, therefore, utilized in a study whose results are being reported here to provide the proper environment required to thoroughly observe the exact effect that visual change to object size had on programming fingertip forces when objects were lifted with a precision grip. Twenty-one participants performed repeated lifts of an identical grip apparatus to a height of 20 mm, maintained each lift for 8 seconds, and then replaced the grip apparatus on the table. While all other factors of the grip apparatus remained unchanged, visual appearance was altered graphically in a 3-D augmented environment. The grip apparatus measured grip and load forces independently. Grip and load forces demonstrated significant rates of increase as well as peak forces as the size of graphical images increased; an aspect that occurred in spite of the fact that extraneous haptic information remained constant throughout the trials. By indicating a human tendency to rely - even unconsciously - on visual input to program the forces in the initial lifting phase, this finding provides further confirmation of previous research findings obtained in the physical environment; including the possibility of extraneous haptic effects (Gordon et al. 1991a, Mon-Williams and Murray 2000, Kawai et al. 2000). The present results also suggest that existing knowledge concerning human manipulation tasks in the physical world may be applied to an augmented environment where the physical objects are enhanced by computer generated visual components.     

从活性顺序图到时态逻辑的转化方法

为了将活性顺序图用于模型检测,方便描述系统的场景需求,提出了一种将活性顺序图转换成时态逻辑的转化方法.分析活性顺序图语言并且定义一种基于路径的语义,用活性顺序图表述系统的场景需求.根据提出的语义,给出了一个将场景需求显式转化为时态逻辑的一般方法,针对并发消息较多的系统扩展和优化此方法,以得到更简短的时态逻辑公式.通过实例说明活性顺序图到线性时态逻辑的转化过程.     

Transfer of content-aware vignetting effect from paintings to photographs

This paper discusses how the vignetting effect of paintings may be transferred to photographs, with attention to center-corner contrast. First, the lightness distribution of both are analyzed. The results show that the painter’s vignette is more complex than that achieved using common digital post-processing methods. It is shown to involve both the 2D and 3D geometry of the scene. Then, an algorithm is developed to transfer the vignetting effect from an example painting to a photograph. The example painting is selected as that has similar contextual geometry with the photograph. The lightness weighting pattern extracted from the selected example painting is adaptively blended with the input photograph to create vignetting effect. In order to avoid over-brightened or over-darkened regions in the enhancement result, the extracted lightness weighting pattern is corrected using a nonlinear curve. A content-aware interpolation method is also proposed to warp the lightness weighting to fit the contextual structure of the photograph. Finally, the local contrast is restored. Experiments show that the proposed algorithm can successfully perform this function. The resulting vignetting effect is more naturally presented with regard to esthetic composition as compared with vignetting achieved with popular software tools and camera models.

     

Das Momentum eines Projekts

Ohne Zusammenfassung     

Toward a Natural Language Interface for Transferring Grasping Skills to Robots

In this paper, we report on the findings of a human-robot interaction study that aims at developing a communication language for transferring grasping skills from a nontechnical user to a robot. Participants with different backgrounds and education levels were asked to command a five-degree-of-freedom human-scale robot arm to grasp five small everyday objects. They were allowed to use either commands from an existing command set or develop their own equivalent natural language instructions. The study revealed several important findings. First, individual participants were more inclined to use simple, familiar commands than more powerful ones. In most cases, once a set of instructions was found to accomplish the grasping task, few participants deviated from that set. In addition, we also found that the participant's background does appear to play a role during the interaction process. Overall, participants with less technical backgrounds require more time and more commands on average to complete a grasping task as compared to participants with more technical backgrounds.     

虚拟环境下物体的手抓取

本文分析了手模型及其虚拟环境下物体的操作方法,提出了点接触平面法矢的手握持抓取方法。在此基础上运用虚拟手模型和抓取规则实现了在虚拟环境下对物体的抓取、移动和释放,实例说明能够使虚拟手正确地抓取虚拟环境中的物体。     

Dynamic Manipulability of Multifingered Grasping

In this paper, we extend the concept of dynamic manipulability to evaluate the dynamic property of multifingered grasping systems consisting of a multifingered hand and a grasped object, and propose a measure of dynamic manipulability of multifingered grasping. Similar to the original dynamic manipulability, the proposed measure evaluates the mapping from a set of realizable joint torques to a set of resultant accelerations of the grasped object, which forms an ellipsoid under a constant internal force constraint. It is clearly shown that the internal forces not only affect the volume of the ellipsoid, but also the amount of offset of the ellipsoid, while the gravity forces simply induce an offset. A new measure, i.e., omnidirectionality, is introduced to add a penalty to the original manipulability measure, which simply evaluates the volume of the ellipsoid, depending on how much the ellipsoid is offset. Numerical examples by using a simple two-fingered robot hand are shown to demonstrate the effectiveness of the proposed measure.