基于碰撞检测的护理型操作臂的安全性设计与实现

针对与人近距离交互的护理型操作臂的安全性问题,提出并实现了一种基于碰撞检测的安全性设计方法,它根据由动力学模型计算获得的参考输出力矩与力矩传感器测得的实际输出力矩间的偏差实现碰撞检测.在关节力矩传感器设计中,采用有限元分析方法来优化应变片的位置布置,并开发了高抗干扰能力的信号处理电路.根据护理型操作臂的低速特点,提出了...     

Aerial cooperative transporting and assembling control using multiple quadrotor–manipulator systems

In this paper, a fully distributed control scheme for aerial cooperative transporting and assembling is proposed using multiple quadrotor–manipulator systems with each quadrotor equipped with a robotic manipulator. First, the kinematic and dynamic models of a quadrotor with multi-Degree of Freedom (DOF) robotic manipulator are established together using Euler–Lagrange equations. Based on the aggregated dynamic model, the control scheme consisting of position controller, attitude controller and manipulator controller is presented. Regarding cooperative transporting and assembling, multiple quadrotor–manipulator systems should be able to form a desired formation without collision among quadrotors from any initial position. The desired formation is achieved by the distributed position controller and attitude controller, while the collision avoidance is guaranteed by an artificial potential function method. Then, the transporting and assembling tasks request the manipulators to reach the desired angles cooperatively, which is achieved by the distributed manipulator controller. The overall stability of the closed-loop system is proven by a Lyapunov method and Matrosov's theorem. In the end, the proposed control scheme is simplified for the real application and then validated by two formation flying missions of four quadrotors with 2-DOF manipulators.     

Adaptive virtual power-based collision detection and isolation with link parameter estimation

ABSTRACT

We propose an adaptive virtual power-based collision detection and isolation approach for robotic manipulators with link parameter estimation. The power indexes are obtained using estimated virtual velocities and contact forces. The effectiveness of the power indexes is relied on precise knowledge of link parameters, which is commonly difficult to obtain. Therefore, in this paper, we propose a series of adaptive power indexes using a link parameter estimation scheme, in order to increase the robustness to parameter uncertainties. To show the statistical performance of collision detection and isolation using the proposed approach, we conduct multiple contact tasks using a 2 degree of freedom (DOF) experimental manipulator while considering uncertainties of the link parameters, and most collisions occurred on each link of the 2-DOF manipulator can be correctly detected and isolated. Additionally, the model-based and adaptive power indexes are compared for collision detection and collision isolation, respectively. Using the adaptive power indexes, the unique threshold can be smoothly determined for collision detection; meanwhile, the correct collision isolation rate increases.     

Fault detection and isolation in robotic manipulators via neural networks: A comparison among three architectures for residual analysis

In this article we discuss artificial neural networks‐based fault detection and isolation (FDI) applications for robotic manipulators. The artificial neural networks (ANNs) are used for both residual generation and residual analysis. A multilayer perceptron (MLP) is employed to reproduce the dynamics of the robotic manipulator. Its outputs are compared with actual position and velocity measurements, generating the so‐called residual vector. The residuals, when properly analyzed, provides an indication of the status of the robot (normal or faulty operation). Three ANNs architectures are employed in the residual analysis. The first is a radial basis function network (RBFN) which uses the residuals of position and velocity to perform fault identification. The second is again an RBFN, except that it uses only the velocity residuals. The third is an MLP which also performs fault identification utilizing only the velocity residuals. The MLP is trained with the classical back‐propagation algorithm and the RBFN is trained with a Kohonen self‐organizing map (KSOM). We validate the concepts discussed in a thorough simulation study of a Puma 560 and with experimental results with a 3‐joint planar manipulator. © 2001 John Wiley & Sons, Inc.     

Control and simulation for a closed-chain dual redundant ' manipulator system

The two-level hierarchical control scheme is adopted to solve a dual-chain robotic system formed by two redundant manipulators grasping a common object with hard point contacts. The upper-level coordinator gathers all the necessary information to resolve the force distribution problem so as to generate all the desired contact forces With all the desired contact forces being specified, the dynamics of each chain are decoupled. Therefore, the lower-level local control problem can be treated as a general open-chain redundant manipulator problem. Furthermore, the Compact-Dual LP method is applied to resolve the upper-level coordination problem; while the Compact QP method associated with the computed torque control method is adopted to solve the redundant manipulator problem. To obtain proper simulation results, the simulated actual contact forces are formulated and the corresponding simulation problem of the closed-chain robotic system is also completely solved in the article. Simulation results show that the two-level hierarchical control scheme is an effective and efficient algorithm for resolving the large-scale control problem of multiple-chain robotic systems. © 1995 John Wiley & Sons, Inc.     

Cooperative object manipulation with contact impact using multiple impedance control

Impedance Control imposes a desired behavior on a single manipulator interacting with its environment. The Multiple Impedance Control (MIC) enforces a designated impedance on both a manipulated object, and all cooperating manipulators. Similar to the standard impedance control, one of the benefits of this algorithm is the ability to perform both free motions and contact tasks without switching control modes. At the same time, the potentially large object inertia and other forces are taken into account. In this paper, the general formulation for the MIC algorithm is developed for distinct cooperating manipulators, and important issues are detailed. Using a benchmark system, the response of the MIC algorithm is compared to that of the Object Impedance Control (OIC). It is shown that in the presence of flexibility, the MIC algorithm results in an improved performance. Next, a system of two cooperating two-link manipulators is simulated, in which a Remote Centre Compliance is attached to the second end-effector. As simulation results show, the response of the MIC algorithm is smooth, even in the presence of an impact due to collision with an obstacle. It is revealed by both error analysis and simulation that under the MIC law, all participating manipulators, and the manipulated object exhibit the same designated impedance behavior. This guarantees good tracking of manipulators and the object based on the chosen impedance laws which describe desired error dynamics, in performing a manipulation task.     

A force-sensorless approach to collision detection based on virtual powers

A collision detection approach for torque-controlled manipulators is proposed to detect and isolate a collision in an unknown environment without using external sensors. A set of artificial indexes physically representing instantaneous powers are introduced. A contact link can be detected and isolated by finding the smallest power index based on full knowledge of the link parameters. However, it is difficult to precisely obtain the link parameters. Therefore, we propose a robust way in which the collision detection and isolation processes are separated. A collision can be detected by comparing a power-based index with a unique threshold, and a contact link can be isolated by comparing power indexes without any additional threshold. The statistical simulation results using a 6 degree of freedom (DOF) spatial manipulator show the performance of the proposed approach in an ideal situation. Furthermore, the statistical experimental results using the 2- and 3-DOF planar manipulators validate the robustness of the proposed approach.     

Neural network approach to trajectory synthesis for robotic manipulators

The paper deals with the collision free trajectory synthesis for industrial robotic manipulators. A new efficient method is proposed that is based on a neural network collision model. The developed iterative transformation procedure provides small computing times for the C-space synthesis and yields sufficiently precise configuration space map for the manipulators with many degrees of freedom. A topologically ordered neural network model is proposed to find the path in the configuration space. The stability of this model is proved using the Lyapunov function technique. To generate the collision model, a modification of the Radial Basis Function Network (RBFN) is used. The developed technique is illustrated by an application example of designing a robotic manufacturing cell for the automotive industry.     

Assembly task execution using visual 3D surface reconstruction: An integrated approach to parts mating

This paper is focused on assembly tasks executed by an industrial robotic manipulator in the presence of uncertainties. The goal is to achieve higher levels of autonomy and flexibility of robotic systems in the execution of such tasks. In particular, as a well-established paradigm of assembly tasks, a Peg-in-Hole task has been considered, where the pose of the target object with respect to the robot is known with uncertainties far larger than the task tolerance, e.g., due to manual positioning of the object in the workcell. The proposed approach is based on the reconstruction of the object surface by means of a number of point clouds provided by a depth sensor. The reconstruction is then compared with a known CAD model of the surface, in order to localize the position and tilt of the holes. Finally, the peg insertion is performed in two steps: a search phase, in which the peg tip gently slides on the surface following a trajectory described by Lissajous functions, and a mechanical coupling phase, in which a compliant behavior is imposed to the peg. Experiments on a collaborative manipulator confirm that the proposed approach allows to achieve a better degree of autonomy and flexibility for a class of robotic tasks in partially structured environments.     

On the Validation of SPDM Task Verification Facility

This paper describes a methodology for validating a ground‐based, hardware‐in‐the‐loop, space‐robot simulation facility. This facility, called “SPDM task verification facility,” is being developed by the Canadian Space Agency for the purpose of verifying the contact dynamics performance of the special purpose dexterous manipulator (SPDM) performing various maintenance tasks on the International Space Station because the real SPDM cannot be physically tested for 3D operations on the ground due to the gravity. The facility uses a high‐fidelity SPDM mathematical model, known as the “truth model” of the space robot, to drive a hydraulic robot to mimic the space robot performing contact operations. In this research different techniques were studied for practically verifying that the complex simulation facility preserves the dynamics of the truth model of the space robot for space‐representative contact robotic tasks. Based upon the study and many years of experience in developing and verifying space robotic systems, a practical validation strategy including detailed test cases was developed along with a set of quantitative criteria for judging the validation test results. © 2004 Wiley Periodicals, Inc.     

A force–impedance controlled industrial robot using an active robotic auxiliary device

A strategy to improve the performance of current commercial industrial robots is presented in this paper. This strategy involves cooperation of two robotic manipulators: the robotic controlled impedance device (RCID) and a commercial industrial robot. The RCID is a small six degrees-of-freedom (DOF) high bandwidth force–impedance controlled parallel manipulator, developed at the School of Engineering of the University of Porto (Portugal). The RCID works attached in series with a position controlled commercial industrial robot. Combination of the two manipulators behaves as a single manipulator having the impedance and force control dynamic performance of the RCID, as well as the workspace and trajectory tracking bandwidth of the industrial robot. Force–impedance control of the RCID, and experimental results on typical tasks that involve end-effector contact with uncertain environments of unknown stiffness are presented.     

Stability and control aspects of flexible link robot manipulators during constrained motion tasks

The effect of robotic manipulator structural compliance on system stability and trajectory tracking performance and the compensation of this structural compliance has been the subject of a number of publications for the case of robotic manipulator noncontact task execution. The subject of this article is the examination of dynamics and stability issues of a robotic manipulator modeled with link structural flexibility during execution of a task that requires the robot tip to contact fixed rigid objects in the work environment. The dynamic behavior of a general n degree of freedom flexible link manipulator is investigated with a previously proposed nonlinear computed torque constrained motion control applied, computed based on the rigid link equations of motion. Through the use of techniques from the theory of singular perturbations, the analysis of the system stability is investigated by examining the stability of the “slow” and “fast” subsystem dynamics. The conditions under which the fast subsystem dynamics exhibit a stable response are examined. It is shown that if certain conditions are satisfied a control based on only the rigid link equations of motion will lead to asymptotic trajectory tracking of the desired generalized position and force trajectories during constrained motion. Experiments reported here have been carried out to investigate the performance of the nonlinear computed torque control law during constrained motion of the manipulator. While based only on the rigid link equations of motion, experimental results confirm that high-frequency structural link modes, exhibited in the response of the robot, are asymptotically stable and do not destabilize the slow subsystem dynamics, leading to asymptotic trajectory tracking of the overall system. © 1992 John Wiley & Sons, Inc.     

基于ai深度学习的机器人碰撞预估计控制器设计

机器人运动过程中与外部障碍物之间容易发生碰撞,当碰撞作用力过大时会造成机器零件损坏的问题,为解决这一问题,设计基于ai深度学习的机器人碰撞预估计控制器。建立人机交互电路与串口通信电路,将伺服电机设备、运动控制器、PC感应装置分别接入既定作用区域内,完成预估计控制器的整体应用结构设计。以PyTorch深度学习框架为基础,定义激活函数,再根据预估计参数的实际取值范围,实现对目标机器人对象的精准检测。按照力矩控制条件表达式,确定碰撞行为的表现强度,完成对机器人运动路径的规划,联合相关应用设备,实现基于ai深度学习的机器人碰撞预估计控制器设计。实验结果表明,ai深度学习算法作用下,机器人与障碍物碰撞部位的接触面积不会超过0.25m2,由碰撞行为导致的外部作用力相对较小,不会造成严重的机器零件损坏问题。     

遥控作业器基于主动时延神经网络的感知和控制

遥控作业器是实现未知或危险环境中作业的有力手段．由于遥控机械手大多工作在不确定的环境中，操作者难以预先知道环境目标的动力学特性，在机械手和环境发生力的交互过程中，需要考虑机械手和环境碰撞时的强非线性对控制回路的影响，因此对机械手和环境碰撞的实时感知和识别是非常重要的．本文利用接近觉传感器测量机械手顶端与环境物体的距离，提出了基于主动时延神经网络的遥控作业器的感知和控制方法．仿真实验表明了该方法的有效性和对接近觉传感器量程的要求     

快速的布料碰撞检测和有效的接触摩擦算法

为了快速处理布料的碰撞检测并获得真实的接触摩擦仿真效果,提出一种基于罚函数的碰撞/接触解决方案.首先,采用质点-弹簧模型进行布料的仿真模拟,在弹簧形变方向添加改进的阻尼力,以减少粒子之间的振荡来保证系统稳定性;其次,采用代数非穿透滤波器对连续碰撞检测算法进行简化求解,快速判断是否存在方程根,提高布料每帧运行的仿真效率;最后,采用库仑约束和接触约束对每个碰撞/接触对进行约束,并结合改进的罚函数法有效地响应所有的碰撞/接触对.实验结果表明,该算法在CPU仿真环境下能快速有效地处理布料的碰撞和接触摩擦,模拟出布料复杂的物理行为,适用于实时的交互应用.     

CPU–GPU Parallel Framework for Real‐Time Interactive Cutting of Adaptive Octree‐Based Deformable Objects

A software framework taking advantage of parallel processing capabilities of CPUs and GPUs is designed for the real‐time interactive cutting simulation of deformable objects. Deformable objects are modelled as voxels connected by links. The voxels are embedded in an octree mesh used for deformation. Cutting is performed by disconnecting links swept by the cutting tool and then adaptively refining octree elements near the cutting tool trajectory. A surface mesh used for visual display is reconstructed from disconnected links using the dual contour method. Spatial hashing of the octree mesh and topology‐aware interpolation of distance field are used for collision. Our framework uses a novel GPU implementation for inter‐object collision and object self collision, while tool‐object collision, cutting and deformation are assigned to CPU, using multiple threads whenever possible. A novel method that splits cutting operations into four independent tasks running in parallel is designed. Our framework also performs data transfers between CPU and GPU simultaneously with other tasks to reduce their impact on performances. Simulation tests show that when compared to three‐threaded CPU implementations, our GPU accelerated collision is 53–160% faster; and the overall simulation frame rate is 47–98% faster.     

机器人碰撞检测方法形式化

为应对更为复杂的任务需求,现代机器人产业发展愈发迅猛.出于协调工作的灵活性、柔顺性以及智能性等多项考虑因素,多臂/多机器人充分发挥了机器人的强大作用,成为现代机器人产业的重要研究热点.在机器人双臂协调运行当中,机械臂之间以及机械臂与外部障碍物之间容易发生碰撞,可能会造成财产损失甚至人员伤亡.对机器人碰撞检测方法进行形式化验证,以球体和胶囊体形式化模型为基础,构建基本几何体单元之间最短距离和机器人碰撞的高阶逻辑模型,证明其相关属性及碰撞条件,建立机器人碰撞检测方法基础定理库,为多机系统碰撞检测算法可靠性与稳定性的验证提供技术支撑和验证框架.     

Performance analysis of an electrohydraulic impedance controller for robotic interaction control

Performance of an electrohydraulic impedance controller, developed in the physical domain, for controlling the contact forces during robotic interaction tasks is analyzed in this paper. The impedance controller is capable of varying the mechanical impedance of the manipulator during an interaction task and thus controls the interaction force. An electrohydraulic servo actuator, appended to a 1-d.o.f. manipulator arm, acts as the controller in the physical domain. The controller performance depends on the physical parameters such as size of the actuator, bulk modulus of the oil, etc., of the hydraulic system. The influence of these physical parameters on the controller performance is analyzed through frequency analysis and acceptable ranges of values for these parameters are determined. Controller sensitivity, stability of the constrained manipulator system and its frequency response are analyzed theoretically and results are presented. The limitation of application of the controller for practical applications is also analyzed and the results are presented.     

基于深度强化学习与旋量法的机械臂路径规划EI北大核心CSCD

深度强化学习在机械臂路径规划的应用中仍面临样本需求量大和获取成本高的问题.针对这些问题,本文基于数据增强的思路,提出了深度强化学习与旋量法的融合算法.本算法通过旋量法将与环境交互所得的自然轨迹进行有效复制,使深度强化学习样本利用率和算法训练效率得到提高;复制轨迹的同时对被控物体、障碍物等环境元素进行同步复制,以此提高机械臂在非结构环境中的泛化性能.最后,在具备物理模拟引擎的Mujoco仿真平台中,通过Fetch机械臂和UR5机械臂在非结构化环境下进行实验对比分析,结果表明了本文算法对于提升深度强化学习样本利用率和机械臂模型泛化性能的可行性及有效性.     

Stabilizing position/force control of robots interacting with environment by learning connectionist structures

This paper is mostly concerned with the application of connectionist architectures for fast on-line learning of robot dynamic uncertainties used at the executive hierarchical control level in robot contact tasks. The connectionist structures are integrated in non-learning control laws for contact tasks which enable stabilization and good tracking performance of position and force. It has been shown that the problem of tracking a specified reference trajectory and specified force profile with a present quality of their transient response can be efficiently solved by means of the application of a four-layer perceptron. A four-layer perceptron is part of a hybrid learning control algorithm through the process of synchronous training which uses fast learning rules and available sensor information in order to improve robotic performance progressively in the minimum possible number of learning epochs. Some simulation results of the deburring process with robot MANUTEC r3 are shown to verify effectiveness of the proposed control learning algorithms.