Content uploaded by Jose Manuel Ortiz-Rodriguez
Author content
All content in this area was uploaded by Jose Manuel Ortiz-Rodriguez on May 23, 2022
Content may be subject to copyright.
A novel optimization robust design of artificial neural networks to solve the inverse kinematics of a manipulator of 6 DOF
... The dataset was composed of 10,000 with 75% used for training, 15% for testing, and 15% for ANN model validation. The model was completed using the ANN algorithm [36][37][38][39][40][41]. In the presen four neural networks were produced sequentially from the robot base, as shown i 10. ...
... The loss function was formulated as follows in Equation (23): The model was completed using the ANN algorithm [36][37][38][39][40][41]. In the present work, four neural networks were produced sequentially from the robot base, as shown in Table 10. ...
Robot acceptance is rapidly increasing in many different industrial applications. The advancement of production systems and machines requires addressing the productivity complexity and flexibility of current manufacturing processes in quasi-real time. Nowadays, robot placement is still achieved via industrial practices based on the expertise of the workers and technicians, with the adoption of offline expensive software that demands time-consuming simulations, detailed time-and-motion mapping activities, and high competencies. Current challenges have been addressed mainly via path planning or robot-to-workpiece location optimization. Numerous solutions, from analytical to physical-based and data-driven formulation, have been discussed in the literature to solve these challenges. In this context, the machine learning approach has proven its superior performance. Nevertheless, the industrial environment is complex to model, generating extra training effort and making the learning procedure, in some cases, inefficient. The industrial problems concern workstation productivity; path-constrained minimal-time motions, considering the actuator’s torque limits; followed by robot vibration and the reduction in its accuracy and lifetime. This paper presents a procedure to find the robot base location for a prescribed task within the robot’s workspace, complying with multiple criteria. The proposed hybrid procedure includes analytical, physical-based, and data-driven modeling to solve the optimization problem. The contribution of the algorithm, for a given user-defined task, is the search for the best robot base location that enables the target points, maximizing the manipulability, avoiding singularities, and minimizing energy consumption. Firstly, the established method was verified using an anthropomorphic robot that considers different levels of a priori kinematics and system dynamics knowledge. The feasibility of the proposed method was evaluated through various simulations for small- and medium-sized robots. Then, a commercial offline program was compared, considering three scenarios and fourteen robots demonstrating an energy reduction in the 7.6–13.2% range. Moreover, the unknown joint dependency in real robot applications was investigated. From 11 robot positions for each active joint, a direct kinematic was appraised with an automatic DH scheme that generates the 3D workspace with an RMSE lower than 65.0 µm. Then, the inverse kinematic was computed using an ANN technique tuned with a genetic algorithm showing an RMSE in an S-shape task close to 702.0 µm. Finally, three experimental campaigns were performed with a set of tasks, repetitions, end-effector velocity, and payloads. The energy consumption reduction was observed in the 12.7–22.9% range. Consequently, the proposed procedure supports the reduction in workstation setup time and energy saving during industrial operations.
... The mechanical components were designed based on the modification of an open-source, six-degree-of-freedom Thor Robot, which was released in 2017 [12]. Since then, the model has undergone many modifications, adaptations, and applications by numerous researchers worldwide and more than 20 units have been built in at least 11 different countries [13][14][15][16][17][18][19]. FreeCAD, an open-source 3D modeling software was used to model the robotic arm. ...
The study presents the development of an accessible, reliable, 3D printable, low-cost, and modular 4 degrees-of-freedom robotic arm for the automated sorting of plastic bottles from the waste stream. The UIArm I robot arm was designed based on the modification of an open-source Thor Robot model using Free-CAD with the components 3D printed using PLA and PETG. The forward kinematics was obtained by Denavit-Hartenberg (DH) method, while the analytical method was used for the inverse kinematics. The electrical components include stepper motors, servo motors, motor drivers, a printed circuit board (PCB), an Arduino Mega microprocessor, a light source for illumination, and a PC with a webcam. Python was used for programming the PC and C# for the Arduino microprocessor. TensorFlow, an end-to-end open-source, machine learning platform was used to develop the object detection algorithm based on a deep neural network. The object detection model achieved an accuracy of 91% for Pepsi plastic bottles which formed the bulk of training images. Other types of plastic bottles were detected with an 85% accuracy. The study has demonstrated the viability of a locally developed robotic arm for the automated sorting of plastic bottles.
... The training can be affected by many factors, but most of the causalities have not been discussed yet. Only a few studies had focused on the NN's hyperparameters design and optimization strategies in this area [7]. It is infrequent in comparison to other application areas of NN. ...
This paper provides guidelines for training a neural network (NN) to encode the mass matrices of articulated rigid-body systems in Cholesky-decomposed form. To store a mass matrix in Cholesky-decomposed form via an NN is expectedly computationally efficient for applications like control and simulation. It is known that training an NN to approximate the mass matrix can be affected by many factors, but most of the causalities have not been discussed yet, and the training takes a very long time. This paper discusses how different training strategies affect the training results. We also propose a norm-based cost function, and we investigate how the order of the norm of the cost function influences the training result. The results show that the NN is trained fast and becomes accurate in different scenarios. NNs achieved approximation error less than 1% during the testing stage in the 2-DOF system after 8 minutes of training; approximation error less than 1% during the testing stage of the 3-DOF system after 58 minutes of training; approximation error less than 2% in the 4-DOF systems after 11 hours and 3 minutes, and 14 hours and 32 minutes of training.
The rapid development of artificial intelligence technology makes Robotics more intelligent and flexible. In this context, the kinematics inverse algorithm of the Robotics has become the basis and key technology for the further development of the Robotics. The key point of the inverse algorithm of the Robotics is to coordinate the Robotics operation arm with the corresponding action actuator at the end to realize the space attitude control of the Robotics system, and to make a theoretical basis for the motion analysis of the later Robotics. However, the traditional form of Robotics kinematics inverse algorithm avoids a lot of iterative computational solution process, which increases the complexity of the whole algorithm. Therefore, based on the above situation, this paper proposes a Robotics inverse solution algorithm based on improved BP (back propagation) neural network. In this paper, in the application of the actual algorithm, aiming at the convergence problem of the traditional BP neural network algorithm, an improved BP neural network algorithm based on the excitation function is proposed. By selecting the adaptive processing function in each layer of the neural network, the selection is matched with it. The learning rate, thus improving the accuracy of the entire motion inverse algorithm. At the same time, in order to further reduce the calculation of joint quantification, this paper also creatively introduces the algorithm of plane division auxiliary dynamic model construction. The simulation results show that the inverse kinematics algorithm based on improved BP neural network proposed in this paper has obvious advantages in solving the kinematics inverse problem of six-degree-of-freedom Robotics compared with the traditional inverse solution algorithm.
A novel spatial parallel manipulator designed to assemble diagnostic instruments in SG-III is introduced in this paper. Firstly, resorting to screw theory, mobility analysis is presented for this manipulator. Then, the inverse kinematics problem is determined by the method of RPY transformation with the singularity analyzed. As a key issue in parallel manipulators, it is more difficult to solve the forward kinematics problem, since it is highly nonlinear and coupled. In this work, three different approaches are presented to deal with this issue, namely, the back propagation neural network, the simplified ant colony optimization, and the proposed improved Newton iterative method. Simulation of each approach is conducted, and their merits and demerits are compared in detail. It is concluded that the improved Newton iterative method, which can provide good initial iteration values, shows the best performance in estimation of the nonlinear forward kinematic mapping of the considered parallel manipulator.
In order to overcome the complexity in solving the inverse kinematics calculation of 7-DOF serial manipulator, a new approach CPABC based on artificial bee colony (ABC) algorithm is proposed. CPABC uses the chaotic mapping to optimize the population distribution of the initial food sources to get rid of the local optimization. The whole group of food sources in CPABC is divided into several subgroups which evolve independently and communicate with each other at a certain frequency to improve the convergence rate. To balance the global and local exploitation, two control parameters are introduced to adjust the search step and the change frequency of the optimization parameter when searching the new food source. CPABC is applied to the inverse kinematics calculation of 7-DOF serial manipulator. the simulation results show that CPABC has stronger global searching ability and more fast convergence rate than that of other ABC algorithms.
The problem of inverse kinematics is fundamental in robot control. Many traditional inverse kinematics solutions, such as geometry, iteration, and algebraic methods, are inadequate in high-speed solutions and accurate positioning. In recent years, the problem of robot inverse kinematics based on neural networks has received extensive attention, but its precision control is convenient and needs to be improved. This paper studies a particle swarm optimization (PSO) back propagation (BP) neural network algorithm to solve the inverse kinematics problem of a UR3 robot based on six degrees of freedom, overcoming some disadvantages of BP neural networks. The BP neural network improves the convergence precision, convergence speed, and generalization ability. The results show that the position error is solved by the research method with respect to the UR3 robot inverse kinematics with the joint angle less than 0.1 degrees and the output end tool less than 0.1 mm, achieving the required positioning for medical puncture surgery, which demands precise positioning of the robot to less than 1 mm. Aiming at the precise application of the puncturing robot, the preliminary experiment has been conducted and the preliminary results have been obtained, which lays the foundation for the popularization of the robot in the medical field.
This paper presents a novel inverse kinematics solution for robotic arm based on artificial neural network (ANN) architecture. The motion of robotic arm is controlled by the kinematics of ANN. A new artificial neural network approach for inverse kinematics is proposed. The novelty of the proposed ANN is the inclusion of the feedback of current joint angles configuration of robotic arm as well as the desired position and orientation in the input pattern of neural network, while the traditional ANN has only the desired position and orientation of the end effector in the input pattern of neural network. In this paper, a six DOF Denso robotic arm with a gripper is controlled by ANN. The comprehensive experimental results proved the applicability and the efficiency of the proposed approach in robotic motion control. The inclusion of current configuration of joint angles in ANN significantly increased the accuracy of ANN estimation of the joint angles output. The new controller design has advantages over the existing techniques for minimizing the position error in unconventional tasks and increasing the accuracy of ANN in estimation of robot’s joint angles.
In this study, a hybrid intelligent solution system including neural networks, genetic algorithms and simulated annealing has been proposed for the inverse kinematics solution of robotic manipulators. The main purpose of the proposed system is to decrease the end effector error of a neural network based inverse kinematics solution. In the designed hybrid intelligent system, simulated annealing algorithm has been used as a genetic operator to decrease the process time of the genetic algorithm to find the optimum solution. Obtained best solution from the neural network has been included in the initial solution of genetic algorithm with randomly produced solutions. The end effector error has been reduced micrometer levels after the implementation of the hybrid intelligent solution system.
Robot manipulators are playing increasingly significant roles in scientific researches and engineering applications in recent years. Using manipulators to save labors and increase accuracies are becoming common practices in industry. Neural networks, which feature high-speed parallel distributed processing, and can be readily implemented by hardware, have been recognized as a powerful tool for real-time processing and successfully applied widely in various control systems. Particularly, using neural networks for the control of robot manipulators have attracted much attention and various related schemes and methods have been proposed and investigated. In this paper, we make a review of research progress about controlling manipulators by means of neural networks. The problem foundation of manipulator control and the theoretical ideas on using neural network to solve this problem are first analyzed and then the latest progresses on this topic in recent years are described and reviewed in detail. Finally, toward practical applications, some potential directions possibly deserving investigation in controlling manipulators by neural networks are pointed out and discussed.
A symbolic notation devised by Reuleaux to describe mechanisms did not recognize the necessary number of variables needed for complete description. A reconsideration of the problem leads to a symbolic notation which permits the complete description of the kinematic properties of all lower-pair mechanisms by means of equations. The symbolic notation also yields a method for studying lower-pair mechanisms by means of matrix algebra; two examples of application to space mechanisms are given.
Redundancy resolution is a critical problem in the control of robotic manipulators. Recurrent neural networks (RNNs), as inherently parallel processing models for time-sequence processing, are potentially applicable for the motion control of manipulators. However, the development of neural models for high-accuracy and real-time control is a challenging problem. This paper identifies two limitations of the existing RNN solutions for manipulator control, i.e., position error accumulation and the convex restriction on the projection set, and overcomes them by proposing two modified neural network models. Our method allows nonconvex sets for projection operations, and control error does not accumulate over time in the presence of noise. Unlike most works in which RNNs are used to process time sequences, the proposed approach is model-based and training-free, which makes it possible to achieve fast tracking of reference signals with superior robustness and accuracy. Theoretical analysis reveals the global stability of a system under the control of the proposed neural networks. Simulation results confirm the effectiveness of the proposed control method in both the position regulation and tracking control of redundant PUMA 560 manipulators.