Robot Dynamic Identification Research Articles

One-Shot Bipedal Robot Dynamics Identification With a Reservoir-Based RNN

One-Shot Bipedal Robot Dynamics Identification With a Reservoir-Based RNN

Robot dynamic parameter identification with error covariance based on sensitivity of dynamic/static motion and feedforward control

Robot dynamic parameter identification with error covariance based on sensitivity of dynamic/static motion and feedforward control

A Framework for Data Driven Dynamic Modeling of Serial Manipulators

Robot dynamic modeling and parameter identification are essential for many analyses. High-fidelity multi-body dynamics simulators can model the robot’s dynamic behavior, but they can’t identify the robot non-linear dynamic model, which is needed for controller design. This study proposes a three-step machine learning framework for obtaining the dynamic equations of serial manipulators from data. This framework consists of three steps. Initially, A library of candidate functions is first constructed, together with a data set based on robot unforced response. Secondly, the best models that can represent dynamic systems for each candidate function utilizing the training data set are then obtained using the SINDy-PI algorithm and Akaike Information Criterion (AIC). Through the MSE and test data, those best models will be reduced to get the functions that best describe the dynamic system. Finally, the dynamic equations that characterize the system are derived using the SINDy algorithm, including the applied force or torque. The technique was tested on three case studies —double pendulum, two-link of KUKA robot, and two-link of Stanford robot. The framework correctly determines the structure of the dynamic system and simultaneously accurately identified its parameters. The framework was able to deal with an ill-conditioned system of equations that arises for complex robot configuration.

Inertial Parameter Identification in Robotics: A Survey

This work aims at reviewing, analyzing and comparing a range of state-of-the-art approaches to inertial parameter identification in the context of robotics. We introduce “BIRDy (Benchmark for Identification of Robot Dynamics)”, an open-source Matlab toolbox, allowing a systematic and formal performance assessment of the considered identification algorithms on either simulated or real serial robot manipulators. Seventeen of the most widely used approaches found in the scientific literature are implemented and compared to each other, namely: the Inverse Dynamic Identification Model with Ordinary, Weighted, Iteratively Reweighted and Total Least-Squares (IDIM-OLS, -WLS, -IRLS, -TLS); the Instrumental Variables method (IDIM-IV), the Maximum Likelihood (ML) method; the Direct and Inverse Dynamic Identification Model approach (DIDIM); the Closed-Loop Output Error (CLOE) method; the Closed-Loop Input Error (CLIE) method; the Direct Dynamic Identification Model with Nonlinear Kalman Filtering (DDIM-NKF), the Adaline Neural Network (AdaNN), the Hopfield-Tank Recurrent Neural Network (HTRNN) and eventually a set of Physically Consistent (PC-) methods allowing the enforcement of parameter physicality using Semi-Definite Programming, namely the PC-IDIM-OLS, -WLS, -IRLS, PC-IDIM-IV, and PC-DIDIM. BIRDy is robot-agnostic and features a complete inertial parameter identification pipeline, from the generation of symbolic kinematic and dynamic models to the identification process itself. This includes functionalities for excitation trajectory computation as well as the collection and pre-processing of experiment data. In this work, the proposed methods are first evaluated in simulation, following a Monte Carlo scheme on models of the 6-DoF TX40 and RV2SQ industrial manipulators, before being tested on the real robot platforms. The robustness, precision, computational efficiency and context of application the different methods are investigated and discussed.

Robot Dynamic Parameter Identification with Error Covariance and Periodic Motion Control based on Sensitivity Analysis

Robot Dynamic Parameter Identification with Error Covariance and Periodic Motion Control based on Sensitivity Analysis

Omnidirectional Mobile Robot Dynamic Model Identification by NARX Neural Network and Stability Analysis Using the APLF Method

In this paper, the NARX neural network system is used to identify the complex dynamics model of omnidirectional mobile robot while rotating with moving, and analyze its stability. When the mobile robot model rotates and moves at the same time, the dynamic model of the mobile robot is complex and there is motion coupling. The change of the model in different states is a kind of symmetry. In order to solve the problem that there is a big difference between the mechanism modeling motion simulation and the actual data, the dynamic model identification of mobile robot in special state based on NARX neural network is proposed, and the stability analysis method is given. To verify that the dynamic model of NARX identification is consistent with that of the mobile robot, the Activation Path-Dependent Lyapunov Function (APLF) algorithm is used to distinguish the NARX neural network model expressed by LDI. However, the APLF method needs to calculate a large number of LMIs in practice and takes a lot of time, and, to solve this problem, an optimized APLF method is proposed. The experimental results verify the effectiveness of the theoretical method.

System Identification of Two-Wheeled Robot Dynamics Using Neural Networks

A system identification of two-wheeled robot (TWR) moving on planar space is presented by applying using neural networks. The system identification is to model the TWR dynamics which is a nonlinear system. The model is applied for estimating the TWR posture during the movements. Neural networks applied in the system identification is multi later perceptron. The neural networks consists of three layers with eight neurons at the first layer, five neurons at the second layer, and three neurons at the third layers. The neural networks is trained to model the TWR dynamics based on a set of input and output data. The system identification is demonstrated through computer simulations. The results show that the system identification using neural networks is able to model the TWR dynamics. The neural networks with learning rate 0.005 is able to estimate the TWR posture with convergence time 0.5 seconds.

Unified Neural Adaptive Control for Multiple Human–Robot–Environment Interactions

To go from human demonstration to robot independent operation, there are generally three phases of interaction to undergo, including human-robot interaction (HRI), human-robot-environment interaction (HREI), and robot-environment interaction (REI). Most existing methods address problems of a single stage. In this article, a unified neural adaptive control method that organically fuses multiple interactions is proposed. HRI, REI, and their coupling effects in HREI are comprehensively considered. First, the iterative least squares method is used for robot dynamics identification based on the linearized momentum observer. The accuracy of external force observation is improved to deal with dynamic uncertainties. The human force and the environmental force are achieved and decoupled by using only a force sensor, a momentum observer, and a selection matrix S. Next, the neural adaptive control method compensating position errors caused by the model uncertainty is addressed. The control system is proved to be stable based on the Lyapunov theorem. The trajectory tracking error under the model uncertainties is reduced. Then, the adaptive admittance control method is introduced. The interaction force of HRI is minimized and the interaction force control of REI is realized. Finally, the proposed method is verified by simulations and experiments.

Deep Learning Aided Dynamic Parameter Identification of 6-DOF Robot Manipulators

Generally, structural uncertainty of the robot dynamics system refers to model error caused by parameter identification, unstructured uncertainty is the unmodeled dynamic characteristic. No matter how elaborate modeling methods are used, there always be uncertainty. Therefore, this article applies deep learning for the first time to aid robot dynamic parameter identification of 6 degrees of freedom robot manipulator for compensation of uncertain factors. Firstly, the relatively accurate prediction of torque is obtained by the physical dynamic model using the parameters identification method (errors are less than 10% of the maximum torques). Secondly, we propose a novel deep neural network based approach called Uncertainty Compensation Model (UCM) to compensate the torque error introduced by the uncertainty. The UCM mainly composed by proposed Input Control Module (ICM) and Error Learning Models (ELMs) based on Long-Short-Term Memory and attention mechanism. The proposed ICM, which effectively avoids the unnecessary interference, is used to control valid input for ELMs. The ELMs, consisted by ELM units, concern extracting salient features from sequence data to predict the joint error. Also, this article summarizes the effects of valid input, timestep and attention mechanism on the performance of the UCM. Finally, the verification of parameter identification and torques compensation is carried out by a Universal Robot 5 manipulator. Compared with the prediction torques of physical dynamic model, the proposed UCM has a good ability to capture the friction characteristics and compensate for the error of local maximum torques, which effectively solves the deficiency of the physical dynamics model and improves prediction accuracy (errors are less than 6% of the maximum torques).

Geometric Robot Dynamic Identification: A Convex Programming Approach

Recent work has shed light on the often unreliable performance of constrained least-squares estimation methods for robot mass-inertial parameter identification, particularly for high degree-of-freedom systems subject to noisy and incomplete measurements. Instead, differential geometric identification methods have proven to be significantly more accurate and robust. These methods account for the fact that the mass-inertial parameters reside in a curved Riemannian space, and allow perturbations in the mass-inertial properties to be measured in a coordinate-invariant manner. Yet, a continued drawback of existing geometric methods is that the corresponding optimization problems are inherently nonconvex, have numerous local minima, and are computationally highly intensive to solve. In this paper, we propose a convex formulation under the same coordinate-invariant Riemannian geometric framework that directly addresses these and other deficiencies of the geometric approach. Our convex formulation leads to a globally optimal solution, reduced computations, faster and more reliable convergence, and easy inclusion of additional convex constraints. The main idea behind our approach is an entropic divergence measure that allows for the convex regularization of the inertial parameter identification problem. Extensive experiments with the 3-DoF MIT Cheetah leg, the 7-DoF AMBIDEX tendon-driven arm, and a 16-link articulated human model show markedly improved robustness and generalizability vis-a-vis existing vector space methods while ensuring fast, guaranteed convergence to the global solution.

Robot Dynamic Parameter Identification based on BBO Algorithm

In this paper, a linearized robot dynamics model is first established. The DH parameters and physical constraints of the robot are given. The joint motion trajectory based on Fourier series is designed then. By setting the conditional number of the observation matrix as the objective function, an Biogeography-based optimization algorithm is applied to identify the dynamic parameters of the robot. The physical constraint calibration and predictive torque results are compared with the traditional genetic algorithm to verify the advanced nature of our method.

Inertia Tensor Properties in Robot Dynamics Identification: A Linear Matrix Inequality Approach

Physical feasibility of robot dynamics identification is currently receiving renovated attention from the research community. Inertia tensor inequalities (namely the positive definite property) have been extensively used among other physical constraints to check physical consistency of identification methods. Recently, an extra inequality associated to inertia tensor eigenvalues has been included to check physical consistency, showing that the previous methods were incomplete. In this paper we show that this extra inequality incorporates the positive definite property, being more restrictive. We also include it in the linear matrix inequality framework for robot dynamics identification to obtain fully physical estimates. The relevance of the extra inequality is verified through real 7-DOF robot manipulator experiments.

Adaptive recurrent neural network with Lyapunov stability learning rules for robot dynamic terms identification

In this paper, a recurrent neural network coupled with Kalman filter is proposed to identify dynamic terms of robotic manipulator. By cooperating some inherent characteristics of robot, this network has the capability to individually identify nonlinear terms using Weighted Augmentation Error (WAE). To present the infrastructure of architecture, an adaptive scheme based on the conventional Back Propagation (BP) is firstly driven using the Gradient Descent (GD) method. Additionally, a stable adaptive updating rule is extracted from the discrete time Lyapunov candidate as an approach for the general nonlinear system identification. Then, this approach is applied to the predefined network. To experimentally validate the computational efficiency and control applicability of the proposed method, Adaptive Neural Network Based Inverse Dynamic Control (ANN-Based-IDC) is employed on a laboratory-scaled twin-rotor CE-150 helicopter. This experiment illustrates enhancement of steady-state performance from 2-to-3 times more in compared with simple PID. Moreover, disturbance rejection and robustness tests admit capability of the method for online dynamic identification in the presence of output and dynamic perturbation.

Dynamic Parameter Identification of Overactuated Parallel Robots

Offline robot dynamic identification methods are based on the use of the inverse dynamic identification model (IDIM), which calculates the joint forces/torques (estimated as the product of the known control signal (the input reference of the motor current loop) with the joint drive gains) that are linear in relation to the dynamic parameters, and on the use of the linear least squares technique to calculate the parameters (IDIM-LS technique). However, as actuation-redundant parallel robots are overactuated, their IDIM has an infinity of solution for the force/torque prediction, depending on the value of the desired overconstraint that is a priori unknown in the identification process. As a result, the IDIM cannot be used as it is for such a class of parallel robots. This paper proposes a procedure for the dynamic parameter identification of actuation-redundant parallel robots. The procedure takes advantage of two possible modified formulations for the IDIM of actuation-redundant robots that can be used for identification purpose. The modified IDIM formulations project some or all input torques/forces onto the robot bodies, thus leading to a unique solution of the model that can then be used in the identification process. A systematic and straightforward way to compute these modified IDIM is presented. The identification of the inertial parameters of a planar parallel robot with actuation redundancy, the DualV, is then carried out using these modified IDIM. Experimental results show the validity of the methods.

Global Identification of Joint Drive Gains and Dynamic Parameters of Robots

Off-line robot dynamic identification methods are based on the use of the inverse dynamic identification model (IDIM), which calculates the joint forces/torques that are linear in relation to the dynamic parameters, and on the use of linear least squares technique to calculate the parameters (IDIM-LS technique). The joint forces/torques are calculated as the product of the known control signal (the input reference of the motor current loop) by the joint drive gains. Then it is essential to get accurate values of joint drive gains to get accurate estimation of the motor torques and accurate identification of dynamic parameters. The previous works proposed to identify the gain of one joint at a time using data of each joint separately. This is a sequential procedure which accumulates errors from step to step. To overcome this drawback, this paper proposes a global identification of the drive gains of all joints and the dynamic parameters of all links. They are calculated altogether in a single step using all the data of all joints at the same time. The method is based on the total least squares solution of an overdetermined linear system obtained with the inverse dynamic model calculated with available input reference of the motor current loop and joint position sampled data while the robot is tracking some reference trajectories without load on the robot and some trajectories with a known payload fixed on the robot. The method is experimentally validated on an industrial Stäubli TX-40 robot.

Physical feasibility of robot base inertial parameter identification: A linear matrix inequality approach

Identification of robot dynamics is a key issue in boosting the performance of model-based control techniques, having also a key role in realistic simulation. Robot dynamic parameters have physical meaning, hence parameter estimations must correspond to physically feasible values. Since it is only possible to identify linear combinations of parameters (the base parameters) the physical feasibility of such combinations cannot be directly asserted. In this paper we show that feasibility conditions define a convex set and can be written as a linear matrix inequality (LMI), suitable for semidefinite programming (SDP) techniques. We propose three methods based on LMI–SDP to deal with the feasibility of base parameters. The first method checks if a given base parameter estimation has physical meaning. The second method corrects infeasible estimations, finding the closest feasible base parameters. These two methods can be applied to existing regression techniques. The third method performs parameter identification through ordinary least squares regression constrained to the feasible space, guaranteeing the optimal solution. Experiments with a seven-link robot manipulator are provided, and efficiency and scalability are discussed.

1P1-P01 トルク計測と外力計測によるロボットアームの力学パラメータの同定(動作計画と制御の新展開(1))

This paper shows a comparison of joint based and contact-force based dynamics identification for a 5 degrees of freedom robotic arm. Off-line robot dynamic identification methods are mostly based on using Inverse Dynamics, which calculates the joint torque or contact-force that are linear in relation to dynamic parameters. The estimation of the parameters is carried out using a total least squares or weighted least squares method. However, these methods can identify only the minimal-set of inertial parameters. In this paper we calculate the dynamic parameters in two ways which are joint based and contact-force based and we also calculate the standard parameters closet to a priori consistent dynamic parameters obtained through CAD data given by the robot manufacturer. Then, we compare joint based and contact-force based identification method with experimental data.

Dynamic identification of the Kuka LWR robot using motor torques and joint torque sensors data

Off-line robot dynamic identification methods use the Inverse Dynamic Identification Model (IDIM), which calculates the motor torques that are linear in relation to the dynamic parameters of both links and drive chains, and use linear least squares technique (IDIM-LS technique). For most robots, the only available data are the motor position and the motor torques which are calculated as the product of the known current reference signal by the joint drive gains. Then the accuracy of links parameters may be limited by noise and error modeling in the drive chains. The Kuka LWR robot (industrial version IIWA: Intelligent Industrial Work Assistant) gives the possibility for an industrial robot to investigate this problem using the joint torque sensors data, measured at the output of the harmonic drive geared drive chains, to identify only the links inertial parameters without the errors coming from the drive chains. This paper focuses on the comparison of the accuracy of the identification of the dynamic parameters of the rigid model of the LWR4+ version, which is very popular in robotics research, using measures of the motor positions and the motor currents, or the torque sensors measurements or both side data. This paper is giving a first complete and reliable identified rigid dynamic model of the LWR4+, publicly available for the robotics community. Moreover, this work shows for the first time the strong result that motor torques calculated from motor currents can identify the links inertial parameters with the same accuracy than using joint torque sensors at the output of the joint drive chains.

Global identification of joint drive gains and dynamic parameters of parallel robots

Off-line robot dynamic identification methods are based on the use of the Inverse Dynamic Identification Model (IDIM), which calculates the joint forces/torques (estimated as the product of the known control signal—the input reference of the motor current loop—with the joint drive gains) that are linear in relation to the dynamic parameters, and on the use of linear least squares technique to calculate the parameters (IDIM-LS technique). Most of the papers dealing with the dynamic parameters identification of parallel robots are based on simple models, which take only the dynamics of the moving platform into account. However, for advanced applications such as output force control, in which the robot interaction force with the environment are estimated from the values of the input reference, both identifications of the full robot model and joint drive gains are required to obtain the best results. In this paper a systematic way to derive the full dynamic identification model of parallel robots is proposed in combination with a method that allows the identification of both robot inertial parameters and drive gains. The method is based on the total least squares solution of an over-determined linear system obtained with the inverse dynamic model. This model is calculated with available input reference of the motor current loop and joint position sampled data while the robot is tracking some reference trajectories without load on the robot and some trajectories with a known payload fixed on the robot. The method is experimentally validated on a prototype of parallel robot, the Orthoglide.

Robot Dynamic Identification Based on the Immune Clonal Selection Algorithm

Robot Dynamic Identification Based on the Immune Clonal Selection Algorithm