Inverse Dynamic Model Research Articles

Adaptive compound control based on generalized Bouc-Wen inverse hysteresis modeling in piezoelectric actuators.

This paper focuses on the study of the dynamic hysteresis compensation and control of piezoelectric actuators so as to improve the swing accuracy of the piezoelectric fast steering mirror mechanism in the photoelectric compound-axis control system. Moreover, in view of the rate dependence and asymmetry of piezoelectric hysteresis, and the complex inversion process of the generalized Bouc-Wen hysteresis model, the Hammerstein dynamic inverse hysteresis model of the piezoelectric actuator is established. To be specific, the static nonlinearity and rate dependence of the piezoelectric inverse hysteresis are represented by the generalized Bouc-Wen inverse model and the auto-regressive exogenous model, respectively, and the parameters of the model are identified by the adaptive beetle swarm optimization algorithm. In the process of the open-loop feedforward compensation, the dynamic positioning accuracy of the piezoelectric actuator is greatly affected by various disturbances and the uncertainty of the hysteresis compensation model. In this context, a compound control strategy that combines the feedforward compensation with the single-neuron adaptive proportion-integration-differentiation control is proposed based on the Hammerstein dynamic inverse hysteresis model of the piezoelectric actuator. The experimental results verify the effectiveness and superiority of the proposed control strategy.

CONTROLADORES PID E LQR APLICADOS À DINÂMICA INVERSA DE UM MANIPULADOR 3-GDL

The application in the industrial manipulator robots has grown over the years making production systems increasingly efficient. Within this context, the need for efficient controllers is required to perform the control of these manipulators. In this work the PID controller (Proportional-Integral-Derivative) and LQR (Linear Quadratic Regulator) is presented from the inverse dynamics model of a RPP (Rotational - Prismatic - Prismatic) cylindrical manipulator. The inverse dynamic model which is modeled on Simulink together with a cascaded PID controller is presented. The PID and LQR results are also presented for joint independent and joint dependent control, i.e a controlled PID is used for each joint, controlling the trajectories and speeds at the same time. This paper has as main contributions the development of the manipulator dynamics model and the design of the LQR and PID controllers applied to the inverse dynamics model, which makes the system simpler to control.

Feed-forward control of elastic-joint industrial robot based on hybrid inverse dynamic model

A novel feedforward control method of elastic-joint robot based on hybrid inverse dynamic model is proposed in this paper. The hybrid inverse dynamic model consists of analytical model and data-driven model. Firstly, the inverse dynamic analytical model of elastic-joint robot is established based on Lie group and Lie algebra, which improves the efficiency of modeling and calculation. Then, by coupling the data-driven model with the analytical model, a feed-forward control method based on hybrid inverse dynamics model is proposed. This method can overcome the influence of the inaccuracy of the analytical inverse dynamic model on the control performance, and effectively improve the control accuracy of the robot. The data-driven model is used to compensate for the parameter uncertainties and non-parameter uncertainties of the analytical dynamic model. Finally, the proposed control method is proved to be stable and the multi-domain integrated system model of industrial robot is developed to verify the performance of the control scheme by simulation. The simulation results show that the proposed control method has higher control accuracy than the traditional torque feed-forward control method.

Inverse dynamics modeling of robots based on sparse spectral gaussian process regression

In robot control systems, inverse dynamics calculation of the robotic arm is essential. In this paper, we propose a different inverse dynamics modeling method—sparse spectral Gaussian process regression (SSGPR). Since using Gaussian process regression for robot inverse dynamics modeling is computationally intensive, to address this drawback, it is proposed to apply sparse spectral Gaussian process regression on robot inverse dynamics modeling, the core of which is the sparsification of the GP’s spectrum for the purpose of improving computational efficiency. Experiments show that sparse spectral Gaussian process regression can improve the computational efficiency and outperform other regression methods while ensuring the computational accuracy.

Tracking control of a precision stage with NARX neural network for friction compensation

The objective of this study is to develop a neural network controller for the friction compensation. The purpose models are used as an inverse model of the frictional force and dynamic behaviour of a system. A proportional-integral-derivative (PID) controller and a neural network system architecture are developed for the Nonlinear Autoregressive with Exogenous inputs (NARX) neural network were proposed to control a precision stage. Firstly, a test signal was used to drive the stage then the derived data was used to train a NARX neural network. This neural network model is the inverse dynamic model of the stages and friction force. An architectural approach of NARX showing promising qualities for dynamic system applications, is analysed in this paper. Utilization of this model is as an estimate of the driving force related with the dynamics of the system against displacement, and is then used as a feed-forward controller to compensate for friction errors. Finally, the experimental systems are established and the result shows that the combination of PID and NARX can improve the tracking performance of the precision stage.

Study and application of a new gas pressure inversion model in coal seam while drilling based on directional drilling technology

China is one of the countries most seriously affected by coal and gas outburst disasters. Gas pressure is an important parameter to evaluate the risk of coal and gas outburst. In order to solve the problem existing in the traditional gas pressure measurement method, a dynamic inversion model of gas pressure while drilling is established based on the gas emission rule of coal particle and the gas radial migration equation of borehole wall in this paper. Combined with the directional drilling system and the special gas emission measuring device, the dynamic inversion model is applied to measure the gas pressure in the coal seam. The results show that: (1) The theoretical value of gas pressure obtained from the new model has a small error range by comparing with the actual value of gas pressure, which indicates that the new model has good reliability and applicability; (2) The distribution range of gas pressure value is wide, which shows the non-uniformity of gas occurrence in coal seam. The heterogeneity of gas pressure value shows an approximate normal distribution law in statistics; (3) Influenced by the occurrence of original gas in coal seam, the evolution law of outburst elimination in each area of coal seam also presents the characteristics of non-uniformity. The research results have important practical significance to assist the early warning of coal and gas outburst and to achieve coal and gas co-mining with safety mining.

A comparative study of two inverse dynamic models of 6 degree-of-freedom rotary Stewart-Gough parallel robot

A comparative study of two inverse dynamic models of 6 degree-of-freedom rotary Stewart-Gough parallel robot

Analysis of Learning Algorithms for Multilayer Neural Networks

The modern stage of development of science and technology is characterized by a rapid increase in the complexity of the created technical systems. The management of such systems requires the development of new management methods, since the modification and improvement of traditional management techniques does not always ensure the fulfillment of stringent requirements for management quality indicators. Classical control methods are mainly based on the theory of linear systems, while most real objects are non-linear. The problem of the synthesis of control systems under conditions of uncertainty is currently one of the central problems in the modern theory of automatic control. The complexity of the control object itself, structural, parametric and information uncertainties in the description of the control object, and the complexity of control problems, the multi criteria of optimization problems, the lack of possible analytical solutions, the need to take into account all the properties of disturbances, etc. The solution to this problem requires a search for alternative approaches to the design of control systems, one of which involves the introduction of neural network systems. Neural network control systems are a high-tech direction of control theory and belong to the class of nonlinear dynamic systems. High performance due to parallelization of input information in combination with the ability to train neural networks makes this technology very attractive for creating control devices in automatic systems. Neural networks can be used to build regulating and switching devices, reference, adaptive, nominal and inverse-dynamic models of objects, on the basis of which objects are studied, analysis of the influence of disturbances acting on an object, determination of the optimal control law, search or calculating the optimal program for changing the impact when changing the values of the parameters of the object and the characteristics of the input data. In addition, neural networks can be used to identify objects, predict the state of objects, recognize, cluster, classify, analyze a large amount of data arriving at high speed from a large number of devices and sensors, and the like. The ability to learn according to a given principle of functioning allows creating automated control systems that are optimal in terms of speed, energy consumption, etc. Naturally, in this case, it is possible to implement several principles of functioning and the transition from one to another. They are a universal tool for modeling multidimensional nonlinear objects and finding solutions to ill-posed problems.

Force-based Active Compliance Control of Hydraulic Quadruped Robot

The hydraulically driven quadruped robot has received extensive attention from many scholars due to its high power density and adaptability to unstructured terrain. However, the research on hydraulic quadruped robots based on torque control is not mature enough, especially in the aspect of multi-rigid body dynamics. In this paper, the most commonly used gait trot is selected as the research object. First, the multi-rigid motion equation of the quadruped robot is established by the spin recursion method based on Lie groups. Next, the Lagrange multiplier is used to represent the constraint force to establish the 12-degree-of-freedom inverse dynamics model of the quadruped robot’s stance phase. And the hybrid dynamics method is used to reduce the dimension of the inversion matrix, which simplifies the solution process of the dynamics model. Then, the trajectory of the foot is planned. Through the analysis of the simplified model, it is concluded that the gait cycle and the initial position of the stance phase are important factors affecting the stability of the trot gait. Finally, the controller framework of the quadruped robot is introduced, and the effectiveness of the algorithm designed in this paper is verified through the co-simulation of the trot gait. The co-simulation results show that the inverse dynamics algorithm can be used as the feedforward of the control system, which can greatly reduce the gains of the PD controller; the robot has good compliance and can achieve stable trotting.

Dynamic Modeling and Control of a Parallel Mechanism Used in the Propulsion System of a Biomimetic Underwater Vehicle

Incorporation of parallel mechanisms inside propulsion systems in biomimetic autonomous underwater vehicles (BAUVs) is a novel approach for motion generation. The vehicle to which the studied propulsion system is implemented presents thunniform locomotion, and its thrust depends mainly on the oscillation from its caudal fin. This paper describes the kinematic and dynamic modeling of a 3-DOF spherical 3UCU-1S parallel robotic system to which the caudal fin of a BAUV is attached. Lagrange formalism was employed for inverse dynamic modeling, and its derivation is detailed throughout this paper. Additionally, the implementation of control strategies to compute forces required to actuate limbs to change platform’s flapping frequencies was developed. Four controllers: classic PD, a feedforward plus feedback PD, an adaptive Fuzzy-PD, and a feedforward plus Fuzzy-PD were compared in different simulations. Results showed that augmenting oscillating frequencies (from 0.5 to 5 Hz) increased the complexity of the path tracking task, where the classic control strategy (i.e., PD) was not sufficient, reaching percentage errors above 9%. Control strategies using feedforward terms combined with adaptive feedback techniques reduced tracking error below 2% even during the presence of external disturbances.

Exoskeleton Active Walking Assistance Control Framework Based on Frequency Adaptive Dynamics Movement Primitives.

This paper introduces a novel exoskeleton active walking assistance control framework based on frequency adaptive dynamics movement primitives (FADMPs). The FADMPs proposed in this paper is an online learning and prediction algorithm which is able to online estimate the fundamental frequency of human joint trajectory, learn the shape of joint trajectory and predict the future joint trajectory during walking. The proposed active walking assistance control framework based on FADMPs is a model-based controller which relies on the human joint torque estimation. The assistance torque provided by exoskeleton is estimated by human lower limb inverse dynamics model which is sensitive to the noise in the joint motion trajectory. To estimate a smooth joint torque profile, the joint motion trajectory must be filtered first by a lowpass filter. However, lowpass filter will introduce an inevitable phase delay in the filtered trajectory. Both simulations and experiments in this paper show that the phase delay has a significant effect on the performance of exoskeleton active assistance. The active assistant control framework based on FADMPs aims at improving the performance of active assistance control by compensating the phase delay. Both simulations and experiments on active walking assistance control show that the performance of active assistance control can be further improved when the phase delay in the filtered trajectory is compensated by FADMPs.

Feed-forward Control Based on Dynamic Model Inversion for Hydraulic Shaking Table

Feed-forward Control Based on Dynamic Model Inversion for Hydraulic Shaking Table

Different types of foot orthoses effect on gait mechanics in patients with rheumatoid arthritis

Foot orthoses are a first line conservative treatment for foot impairments in patients with rheumatoid arthritis (RA), however their effect on gait mechanics is poorly understood. We aimed to compare changes in lower limb and foot mechanics between two types of commonly used foot orthoses (FO) with a control. Twenty-seven patients with rheumatoid arthritis participated in this crossover study. Two different types of FO (a medially wedged custom-made FO and a prefabricated FO with a metatarsal dome, respectively), were compared against a control insole. During gait, lower limb mechanics were analyzed using 3D motion capture, force plates, and an in-shoe pressure system. Inverse dynamics models were created in the Anybody Modeling System to calculate joint angles and joint moments during gait. Gait variables were analyzed using statistical parametric mapping. Compared to the control, the prefabricated FO had limited effect on gait mechanics. Compared to the control the custom-made FO reduced ankle plantarflexion moment with 0.4 %body weight * body height (BW * BH) between 66 and 76% of stance and ankle eversion moment was reduced 0.16% BW*BH between 3 and 40% of stance. Furthermore, it also reduced the average forefoot plantar pressure by 9 kPa between 20 and 62% of stance compared to the control. Changes in foot pressure distribution, joint moments and angles were most pronounced for custom-made FO compared to the prefabricated FO. The findings suggest that patients with RA and foot impairments may benefit more from an individualized FO strategy, if the aim of the treatment is to alter gait mechanics. (NCT03561688).

로봇 제어 교육에서 외란 관측기와 시간지연제어기의 연관성에 대한 튜토리얼

This paper presents a unified approach for obtaining the relationship between disturbance observers (DOB) and time-delayed controllers (TDC) for controlling robot manipulators. Previously, DOBs in the linear control configuration have been actively used in motion control applications to reject disturbances to the system. The extended nonlinear DOB can be constructed by formulating inverse dynamics models of robot manipulators. Furthermore, the nonlinear DOB scheme without dynamics models can be constructed for robot manipulators. This scheme leads to the concept of time-delayed control, which means that the two schemes are equivalent. Simulation examples are provided for each scheme. Experimental studies are conducted to compare the performance between proportional-derivative (PD) control and TDC.

Trajectory tracking with collision avoidance for a parallel robot with flexible links

End-effector trajectory tracking of a highly flexible three-link parallel robot is combined with collision avoidance in real-time. Typical real-time collision avoidance algorithms issue commands on actuator level which can cause large oscillations within flexible robots. Thus, the needed evasive motion at the points which are about to collide is used to adapt the desired end-effector trajectory of the underactuated system, even for actuator limits and points other than the end-effector. In contrast to existing real-time collision avoidance concepts, which are essentially completely based on rigid models, a dynamic flexible model inversion is then used for tracking. Thereby, output redefinition leads to a minimum phase system. The presented computationally efficient concepts, being actuator limit and obstacle collision avoidance, use sigmoid functions providing smooth trajectories. The former calculates a braking factor for the end-effector when an actuator approaches its limit. The latter avoids external dynamic obstacles based on repulsive vectors. Experiments validate both concepts with a flexible model inversion which clearly outperforms the classical rigid body approach.

Evaluation of amplitude- and frequency-based techniques for attenuating inertia-based movement artifact during surface translation perturbations

BackgroundSurface translations are a method of perturbing an individual’s balance to evoke balance control responses. However, the force plates used to measure kinetic responses often contain artifacts due to inertial properties coupled with the dynamics of surface translation perturbations. Techniques to attenuate these movement artifacts are not well established within the literature. Research questionAre amplitude- or frequency-based subtraction processing techniques effective at attenuating inertia-based movement artifacts in kinetic signals during surface translations? MethodsOne-hundred and two backward surface translations were analyzed from five participants. Perturbation-matched unloaded pre-trials were collected to characterize force plate movement artifacts. For each trial, baseline data was processed to account for inertial artifacts using both amplitude- and frequency-based subtraction methods producing 3 datasets. Root mean square error (RMSE) between the datasets and expected tracings of an unloaded force plate were calculated. The effects of these processes on calculated knee flexion/extension (FE) moment were characterized using an inverse dynamics model which incorporated ground reaction forces and participant kinematics. ResultsBoth amplitude- and frequency-based processing methods resulted in near identical changes and substantially reduced RMSE values compared to original data. An RMSE reduction of 91.4 % was observed for the unloaded force channel which aligned with the direction of translation. Peak stance knee FE moments decreased by an average of 3.7 Nm and the average largest difference between the original data and both processing techniques across all trials was 30.4 Nm. SignificanceThe results provide quantitative evidence that both data processing methods can successfully attenuate the presence of movement artifact found within force plate signals during surface translations. This study provides recommendations to researchers on effective methods for improving data processing techniques to attenuate force plate movement artifacts introduced by surface translations, towards more accurate estimates of joint kinetics during balance reactions.

The Role of Geometrical Parameters in the Lubrication of the Swiss Lever Escapement

Abstract The lubrication regimes of a contact pair escapement-ruby pallet of a Swiss lever escapement have been investigated combining the theory of fluid lubrication with a well-established kinematic and inverse dynamic multibody model. The kinematic analytical results have been confronted with experimental measurements. The developed model allows to easily obtain, for the three operating phases of the Swiss lever escapement, the relative speed and the contact forces and, by considering a hydrodynamic lubrication regime, the lubricant minimum film thickness and the coefficient of friction. The presented formulation allows to study the influence of crucial technical parameters in the Swiss lever escapement lubrication. The spout radii of curvature have been identified as the optimal parameters to control the lubrication regimes in the pallet/escapement contacts. In that sense, an interesting result is that the lubrication regime moves away from the boundary lubrication by increasing these radii.

Visual Navigation With Multiple Goals Based on Deep Reinforcement Learning.

Learning to adapt to a series of different goals in visual navigation is challenging. In this work, we present a model-embedded actor-critic architecture for the multigoal visual navigation task. To enhance the task cooperation in multigoal learning, we introduce two new designs to the reinforcement learning scheme: inverse dynamics model (InvDM) and multigoal colearning (MgCl). Specifically, InvDM is proposed to capture the navigation-relevant association between state and goal and provide additional training signals to relieve the sparse reward issue. MgCl aims at improving the sample efficiency and supports the agent to learn from unintentional positive experiences. Besides, to further improve the scene generalization capability of the agent, we present an enhanced navigation model that consists of two self-supervised auxiliary task modules. The first module, which is named path closed-loop detection, helps to understand whether the state has been experienced. The second one, namely the state-target matching module, tries to figure out the difference between state and goal. Extensive results on the interactive platform AI2-THOR demonstrate that the agent trained with the proposed method converges faster than state-of-the-art methods while owning good generalization capability. The video demonstration is available at https://vsislab.github.io/mgvn.

Dragondrop: a novel passive mechanism for aerial righting in the dragonfly

Dragonflies perform dramatic aerial manoeuvres when chasing targets but glide for periods during cruising flights. This makes dragonflies a great system to explore the role of passive stabilizing mechanisms that do not compromise manoeuvrability. We challenged dragonflies by dropping them from selected inverted attitudes and collected 6-degrees-of-freedom aerial recovery kinematics via custom motion capture techniques. From these kinematic data, we performed rigid-body inverse dynamics to reconstruct the forces and torques involved in righting behaviour. We found that inverted dragonflies typically recover themselves with the shortest rotation from the initial body inclination. Additionally, they exhibited a strong tendency to pitch-up with their head leading out of the manoeuvre, despite the lower moment of inertia in the roll axis. Surprisingly, anaesthetized dragonflies could also complete aerial righting reliably. Such passive righting disappeared in recently dead dragonflies but could be partially recovered by waxing their wings to the anaesthetised posture. Our kinematics data, inverse dynamics model and wind-tunnel experiments suggest that the dragonfly's long abdomen and wing posture generate a rotational tendency and passive attitude recovery mechanism during falling. This work demonstrates an aerodynamically stable body configuration in a flying insect and raises new questions in sensorimotor control for small flying systems.

Accelerated Convergence Interleaving Iterative Learning Control and Inverse Dynamics Identification

This work aims to quickly identify an FIR inverse dynamical model for linear time-invariant (LTI) systems. Various applications are enabled using the constructed inverse filter, as illustrated by an inversion-based iterative learning control (ILC) algorithm. With the help of interleaving inversion-based ILC and ILC-based inverse dynamics identification, accelerated convergence is obtained. The proposed method removes the numerical instability issues in the calculation of an inverse model. Hence, it is shown more robust against measurement noises. Both simulation comparison and experimental results demonstrate the efficacy and advantages of the proposed strategy.