Artificial neural network dexterous robotics hand optimal control methodology: grasping and manipulation forces optimization

Optimal fingertip forces can always be computed through the well-known optimization algorithms. However, computation time has always remained a real-time constraint. This article presents an efficient scheme to compute optimal grasping and manipulation forces for dexterous robotics hands. This is expressed as a quadratic optimization problem, and an artificial neural network (ANN) is used to learn such quadratic optimization formulations. Computation has been based on a nonlinear model of fingertip contacts and slips. In achieving object grasping while in motion, the hand Jacobian is considered an important matrix to be computed, but it is also highly intensive for real-time computed applications. Consequently, we investigated an efficient approach using artificial neural networks to learn optimal grasping forces. An ANN is used here to learn the optimal contact forces relating hand joint-space torques to the resulting object force. The results have indicated that the ANN has reduced computation times to reasonable values owing to its ability to map nonlinear force relations. Furthermore, the results have revealed that ANNs are capable of learning highly nonlinear relations relating to distributed fingertip forces and joint torques. The technique developed has also proved to be suitable for off-line learning of computed fingertip forces, even with large training samples.     

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%。     

多模式刚柔结合欠驱动抓取机构的设计、分析与实验测试

结合刚性欠驱动抓取机构与柔顺机构,提出一种多模式刚柔结合欠驱动抓取机构,并对所提出的机构进行分析与实验研究。基于刚体替代法,设计了二指多模式欠驱动抓取机构的刚柔结合方案。运用正运动学分析与载荷平衡方程,对驱动单元进行静力学建模。结合伪刚体模型法、载荷平衡方程与操作对象平衡位置枚举搜索,建立两点抓取与包络抓取模式下抓取力与驱动力矩的关系式。将驱动单元静力学建模与抓取单元静力学建模相结合,可以得到完整的多模式抓取力模型。进一步地,考虑由于接触引起的柔性杆件变形,结合线性插值,对抓取力模型进行修正。基于修正后的抓取力模型,对机构尺寸参数进行优化设计,综合提升机构在两点抓取模式和包络抓取模式下的载荷输出性能。RecurDyn仿真结果显示,在两点抓取模式和包络抓取模式下,修正后完整的抓取力模型与仿真值的最大相对误差为7.62%,并且所提出的优化算法有效提升了机构的两点抓取力与综合包络抓取力。实验结果显示,优化后的抓取机构抓取力有较大的提升,修正后完整的抓取力模型与实验值的最大相对误差为1.87%,验证了抓取力建模、优化设计的有效性。     

Comparison of RBF and MLP neural networks to solve inverse kinematic problem for 6R serial robot by a fusion approach

In this paper, a fusion approach to determine inverse kinematics solutions of a six degree of freedom serial robot is proposed. The proposed approach makes use of radial basis function neural network for prediction of incremental joint angles which in turn are transformed into absolute joint angles with the assistance of forward kinematics relations. In this approach, forward kinematics relations of robot are used to obtain the data for training of neural network as well to estimate the deviation of predicted inverse kinematics solution from the desired solution. The effectiveness of the fusion process is shown by comparing the inverse kinematics solutions obtained for an end-effector of industrial robot moving along a specified path with the solutions obtained from conventional neural network approaches as well as iterative technique. The prominent features of the fusion process include the accurate prediction of inverse kinematics solutions with less computational time apart from the generation of training data for neural network with forward kinematics relations of the robot.     

A neuro-simulated annealing approach to the inverse kinematics solution of redundant robotic manipulators

The neural-network-based inverse kinematics solution is one of the recent topics in the robotics because of the fact that many traditional inverse kinematics problem solutions such as geometric, iterative and algebraic are inadequate for redundant robots. However, since the neural networks work with an acceptable error, the error at the end of inverse kinematics learning should be minimized. In this study, simulated annealing (SA) algorithm was used together with the neural-network-based inverse kinematics problem solution robots to minimize the error at the end effector. The solution method is applied to Stanford and Puma 560 six-joint robot models to show the efficiency. The proposed algorithm combines the characteristics of neural network and an optimization technique to obtain the best solution for the critical robotic applications. Three Elman neural networks were trained using separate training sets and different parameters, since one of them can give better results than the others can. The best result is selected within three neural network results by computing the end effector error via direct kinematics equation of the robotic manipulator. The decimal part of the neural network result was improved up to 10 digits using simulated annealing algorithm. The obtained best solution is given to the simulated annealing algorithm to find the best-fitting 10 digits for the decimal part of the solution. The end effector error was reduced significantly.     

Grasping posture learning with noisy sensing information for a large scale of multifingered robotic systems

In this paper, we propose a new method to learn multifingered hand configuration during grasping in the presence of noise and uncertainty. The developed model is composed of two modules. The first one carries out the learning of the fingers inverse kinematics. It is based on a modular architecture composed of several neural networks. Using reinforcement learning, a second neural network based model optimizes the position and orientation of the hand palm taking into account noisy sensing information. Working together these two modules exchange information to define the complete hand configuration. In order to illustrate the capabilities of the proposed model, simulation results are proposed using different kinds of objects, different levels of noise, and for a multifingered hand with different number of fingers. © 2005 Wiley Periodicals, Inc.     

虚拟手抓持规则融合策略研究

群体虚拟手抓持规则是虚拟手和虚拟物体进行抓持操作的交互规则,用于判定虚拟手是否能够成功抓持物体。对基于几何的虚拟手抓持规则和基于物理的虚拟手抓持规则分别进行了研究,针对基于几何的虚拟手抓持规则规则简单、仿真效果较差,基于物理模型的虚拟手抓持规则计算复杂、难以实现实时仿真的问题:（1）改进基于几何的虚拟手抓持规则,通过接触点位置、法矢和抓持面法矢制定抓持规则,使其效果逼近力封闭虚拟手抓持规则;（2）利用力封闭计算中抓持接触点和法矢不变的特性,通过内力配比避免了抓持操作中的非线性规划求解,使抓持操作阶段实现实时仿真;（3）通过几何约束进行初始抓持判断-力封闭计算校正-内力配比力封闭计算的策略,实现了完整的抓持过程实时仿真。设计的交互实验说明该抓持规则能实现高沉浸感和实时性的抓持仿真,可以应用到虚拟训练、虚拟装配等仿真平台。     

A dual neural network for kinematic control of redundant robotmanipulators

The inverse kinematics problem in robotics can be formulated as a time-varying quadratic optimization problem. A new recurrent neural network, called the dual network, is presented in this paper. The proposed neural network is composed of a single layer of neurons, and the number of neurons is equal to the dimensionality of the workspace. The proposed dual network is proven to be globally exponentially stable. The proposed dual network is also shown to be capable of asymptotic tracking for the motion control of kinematically redundant manipulators.     

基于点云采样权重估计的未知物体抓取位姿生成方法

针对机械臂在非结构环境中对未知物体抓取位姿生成困难及抓取稳定性差的问题,提出一种基于点云采样权重估计的抓取位姿生成方法.首先通过移动深度相机的方式拼接得到较完整的物体点云信息,并对物体的几何特性进行分析,有效避开物体不宜抓取的位置进行抓取位姿样本生成;然后结合几何约束条件实现抓取位姿搜索,并利用力封闭条件对样本稳定性进行评估;最后为了对实际的抓取位姿进行评价,根据其稳定性、夹取深度、夹取角度等设定抓取可行性指标,据此在工作空间输出最佳抓取位姿并完成指定的抓取任务.实验结果表明,采用所提方法能够高效生成大量且稳定的抓取位姿,并在仿真环境中有效实现机械臂对单个或多个随机摆放的未知物体的抓取任务.     

基于多约束条件的机器人抓取策略学习方法

针对机器人在多类别物体不同任务下的抓取决策问题,提出基于多约束条件的抓取策略学习方法.该方法以抓取对象特征和抓取任务属性为机器人抓取策略约束,通过映射人类抓取习惯规划抓取模式,并采用物体方向包围盒(OBB)建立机器人抓取规则,建立多约束条件的抓取模型.利用深度径向基(DRBF)网络模型结合减聚类算法(SCM)实现抓取策略的学习,两种算法的结合旨在提高学习鲁棒性与精确性.搭建以Refiex 1型灵巧手和AUBO六自由度机械臂组成的实验平台,对多类别物体进行抓取实验.实验结果表明,所提出方法使机器人有效学习到对多物体不同任务的最优抓取策略,具有良好的抓取决策能力.     

An efficient motion generation method for redundant humanoid robot arms based on motion continuity

This paper proposes a novel method of motion generation for redundant humanoid robot arms, which can efficiently generate continuous collision-free arm motion for the preplanned hand trajectory. The proposed method generates the whole arm motion first and then computes the actuators’ motion, which is different from IK (inverse kinematics)-based motion generation methods. Based on the geometric constraints of the preplanned trajectory and the geometric structure of humanoid robot arms, the wrist trajectory and elbow trajectory can be got first without solving inverse kinematics and forward kinematics. Meanwhile, the constraints restrict all feasible arm configurations to an elbow-circle and reduce the arm configuration space to a two-dimension space. By combining the configuration space and collision distribution of arm motion, collision-free arm configurations can be identified and be used to generate collision-free arm motion, which can avoid unnecessary forward and inverse kinematics. The experiments show that the proposed method can generate continuous and collision-free arm motion for preplanned hand trajectories.     

Control of an object with parallel surfaces by a pair of finger robots without object sensing

This paper proposes a method for controlling an object with parallel surfaces in a horizontal plane by a pair of finger robots. The control method can achieve stable grasping, relative orientation control, and relative position control of the grasped object. The control inputs require neither any object parameters nor any object sensing, such as tactile sensors, force sensors, or visual sensors. The control inputs are also quite simple and do not need to solve either inverse kinematics or inverse dynamics. The stability of the closed-loop system is proved, and simulation and experimental results validate the control method.     

Robot arm reaching through neural inversions and reinforcement learning

We present a neural method that computes the inverse kinematics of any kind of robot manipulators, both redundant and non-redundant. Inverse kinematics solutions are obtained through the inversion of a neural network that has been previously trained to approximate the manipulator forward kinematics. The inversion provides difference vectors in the joint space from difference vectors in the workspace. Our differential inverse kinematics (DIV) approach can be viewed as a neural network implementation of the Jacobian transpose method for arm kinematic control that does not require previous knowledge of the arm forward kinematics. Redundancy can be exploited to obtain a special inverse kinematic solution that meets a particular constraint (e.g. joint limit avoidance) by inverting an additional neural network The usefulness of our DIV approach is further illustrated with sensor-based multilink manipulators that learn collision-free reaching motions in unknown environments. For this task, the neural controller has two modules: a reinforcement-based action generator (AG) and a DIV module that computes goal vectors in the joint space. The actions given by the AG are interpreted with regard to those goal vectors.     

基于卷积神经网络的改进机械臂抓取方法

为了提高机械臂抓取的精度,提出一种基于Mask R-CNN的机械臂抓取最佳位置检测框架。基于RGB-D图像,所提框架通过精确的实例分割确定抓取对象的类别、位置和掩码信息,由反距离加权法在去噪后的深度图上获取中心点的加权深度坐标,构成目标对象的三维目标位置,经坐标系转换得到最终的最优抓取位置。建议的框架考虑到目标对象的姿态与边缘信息,可以有效地提高抓取性能。最后,基于UR3机械臂上的抓取实验结果验证了该框架的有效性。     

基于GA-NN的复杂工艺生产过程多目标优化研究

在复杂非线性多目标优化问题求解中,非线性模型结构很难事先给定,需要检验的参数也非常繁多,应用传统的建模方法和优化模型已难以解决更为复杂的现实问题。人工神经网络技术为解决复杂非线性系统建模问题提供了一条新的途径。将神经网络响应面作为目标函数或者约束条件,加上其他常规约束条件进行系统模型的建立,再应用遗传算法进行优化,从而实现设计分析与设计优化的分离。以某化工企业的生产过程优化问题为例,利用BP神经网络建立了工艺参数与性能目标之间的模型,然后利用遗传算法搜索最优工艺参数,获取了用于指导生产的样本点数据。研究结果表明,该方法能够获得高精度的多目标优化模型,从而使优化效率大为提高。     

A Lagrangian network for kinematic control of redundant robotmanipulators

A recurrent neural network, called the Lagrangian network, is presented for the kinematic control of redundant robot manipulators. The optimal redundancy resolution is determined by the Lagrangian network through real-time solution to the inverse kinematics problem formulated as a quadratic optimization problem. While the signal for a desired velocity of the end-effector is fed into the inputs of the Lagrangian network, it generates the joint velocity vector of the manipulator in its outputs along with the associated Lagrange multipliers. The proposed Lagrangian network is shown to be capable of asymptotic tracking for the motion control of kinematically redundant manipulators.     

Dynamic grasp planning of multifingered robot hands based onasymptotic stability

We describe an approach for planning grasps of multifingered robot hands based on a small vibration model. Using features of the grasp configuration, we analyze asymptotic stability, contact situations, and uniaxial fingertip force constraints for the combined planning of finger posture and finger position, and characterize the generalized mass, damping, and stiffness. Choosing the largest time constant of the vibration model as an optimization criterion for planning finger postures and positions, the original problem of dynamic grasp planning is formulated as a nonlinear program. Simulation examples for a three-fingered robot hand grasping a spherical object demonstrate the effectiveness of the approach.     

改进的自适应多种群DE的机械臂控制方法

根据机械臂关节轴线方向建立了连杆坐标系,利用Denavit-Hartenberg（D-H）法得到连杆坐标系变换矩阵;通过连杆坐标系变换矩阵得到机械臂正运动控制模型;通过正运动模型得到逆运动控制模型,逆运动控制模型是多目标约束优化问题,该模型的最优解既可以保证控制精度,又可以保证各个关节角变动幅度的总代价（定义为旋转副）达到最小。为了求解机械臂逆运动模型,提出了自适应多种群差分演化算法（AMPDE）,多种群策略可以提升个体共享群体信息的能力,自适应变异策略可以提升种群多样性,数值实验表明该算法可以有效求解机械臂逆运动学模型。