Human-Inspired Force-Motion Imitation Learning with Dynamic Response for Adaptive Robotic Manipulation

Magnetic levitation system is a typical nonlinear and instable system. Based on the complexity and inaccuracy of modelling, in this paper identified magnetic levitation system using the speciality that neural network(NN) can approach any nonlinear function. A Radial Basis Function neural network (RBFNN) controller is designed based on the neural network adaptive control principle. This paper proposes a control method which combine neural network adaptive control method and state feedback control method based on RBFNN. A simulation of the system is proposed, and the result shows that RBFNN could approach magnetic levitation system very well, neural network adaptive state feedback controller has a good effect on this nonlinear system; this control system has a preferable stability and control property.

In this paper, composite dynamic movement primitives (DMPs) based on radial basis function neural networks (RBFNNs) are investigated for robots’ skill learning from human demonstrations. The composite DMPs could encode the position and orientation manipulation skills simultaneously for human-to-robot skills transfer. As the robot manipulator is expected to perform tasks in unstructured and uncertain environments, it requires the manipulator to own the adaptive ability to adjust its behaviours to new situations and environments. Since the DMPs can adapt to uncertainties and perturbation, and spatial and temporal scaling, it has been successfully employed for various tasks, such as trajectory planning and obstacle avoidance. However, the existing skill model mainly focuses on position or orientation modelling separately; it is a common constraint in terms of position and orientation simultaneously in practice. Besides, the generalisation of the skill learning model based on DMPs is still hard to deal with dynamic tasks, e.g., reaching a moving target and obstacle avoidance. In this paper, we proposed a composite DMPs-based framework representing position and orientation simultaneously for robot skill acquisition and the neural networks technique is used to train the skill model. The effectiveness of the proposed approach is validated by simulation and experiments.

The autonomous vehicle is an automated system equipped with features like environment perception, decision-making, motion planning, and control and execution technology. Navigating in an unstructured and complex environment is a huge challenge for autonomous vehicles, due to the irregular shape of road, the requirement of real-time planning, and the nonholonomic constraints of vehicle. This paper presents a motion planning method, based on the Radial Basis Function (RBF) neural network, to guide the autonomous vehicle in unstructured environments. The proposed algorithm extracts the drivable region from the perception grid map based on the global path, which is available in the road network. The sample points are randomly selected in the drivable region, and a gradient descent method is used to train the RBF network. The parameters of the motion-planning algorithm are verified through the simulation and experiment. It is observed that the proposed approach produces a flexible, smooth, and safe path that can fit any road shape. The method is implemented on autonomous vehicle and verified against many outdoor scenes; furthermore, a comparison of proposed method with the existing well-known Rapidly-exploring Random Tree (RRT) method is presented. The experimental results show that the proposed method is highly effective in planning the vehicle path and offers better motion quality.

Short-term load forecasting of power system is not only the basis of scheduling of generating sets, but also the basis of working out the transaction schedule in electricity market. This paper proposes a short-term load forecasting method based on combination of radial basis function (RBF) neural network and fuzzy control, uses on-line self-modify factor fuzzy control to eliminate forecast error on the basis of RBF neural network forecasting. The practical examples show that the accuracy of short-term load forecasting and training speed can be improved and gains the very satisfactory results by the proposed method

An adaptive feedback control scheme and a switching strategy for the output tracking stabilization of switched nonlinear systems with time-delay are presented by using RBF (radial basis function) neural network The RBF neural networks is used to compensate adaptively for the unknown nonlinear parts of switched nonlinear systems, and the approximation errors of RBF neural networks are introduced to the adaptive law in order to improve the tracking quality of the systems. The controller and switching strategy can guarantee asymptotic stability of tracking error for the overall system

This article presents an adaptive-adaptive radial basis function (RBF) neural network (AANNRBF) control strategy for lower limb rehabilitation robot (LLRR), which to adapts the uncertainty of the physical parameters caused by different patients or the same patient's muscle tension changes in different training periods. The load disturbance changes and the uncertainty of the system physical parameters generated in the human-machine interaction are controlled by parameter adaptive control, which solves the problem that the neural net¬work approximation domain of the traditional purely adaptive RBF neural network (ANNRBF) control algorithm is easily a¬ffected by the change of the load. ADAMS (automatic dynamic analysis of mechanical system)-MATLAB co-simulation results show that the control method proposed can achieve accurate gait trajectory tracking training, and has good adaptability and robustness to system load fluctuations caused by changes in muscle tension during patient training.

This article proposed a novel control approach for trajectory tracking of robotic airships. First, the nonlinear dynamics model of a robotic airship is derived and formulated. Second, a sliding mode controller is designed to track the commanded trajectory for its robustness against parametric variations and external disturbances. To address the chattering problem results from the sliding mode controller, a radial basis function neural network is adopted to construct the neural network self-gain-scheduling sliding mode controller in which the control gains are scheduled synchronously with the sliding surface via radial basis function neural network according to the learning algorithm, with switching sliding surface and its differential as radial basis function neural network inputs and control gains as radial basis function neural network outputs. In addition, the stability and convergence of the closed-loop controller are proven using the Lyapunov stability theorem. Finally, the effectiveness and robustness of the proposed controller are demonstrated via numerical simulations. Contrasting simulation results indicate that the neural network self-gain-scheduling sliding mode controller attenuates the chattering effectively and has better performance against the sliding mode controller.

In this paper, two novel adaptive finite-time control schemes are proposed for position tracking of nonlinear teleoperation system, which dynamic uncertainties, actuator saturation, and time-varying communication delays are considered. First, a novel auxiliary variable is designed to provide more stable performance. The radial basis function (RBF) neural network is introduced to estimate dynamic uncertainties. Second, two adaptive finite-time control schemes are investigated. In control scheme I, the RBF neural network and the gain switching strategy are applied to compensate the actuator saturation. In control scheme II, an auxiliary compensation filter and the compensation adaptive update laws, which contain the finite-time structure, are developed for dealing with saturation. Third, the finite-time adaptive controller is designed in each of these two control schemes. Based on the multiple Lyapunov function method, the closed-loop teleoperation system with these two control methods is proved to be bounded and finite-time stability. Finally, the simulation experiments are performed and the comparisons with other control methods are shown. The effectiveness of the proposed control schemes is demonstrated.

We propose a radial basis function (RBF) neural network controller for ship control. The main feature of the controller is to combine a fuzzy controller with neural networks. We transform a set of fuzzy inference rules into an RBF neural-network controller, utilizing the nonlinear mapping and learning ability of the neural networks. We present the design method of the controller based on the RBF neural networks and fuzzy rules, and discuss the characteristics of the controller. We use MATLAB as a simulation language. Simulation tests are conducted with different parameters, and the results are given in the paper. We introduce a ship control test bed. The controller is installed on the ship control test bed, and sea trials are conducted in the open sea. The sea trial results are also shown in the paper.

In this article, a novel sliding mode control is proposed for synchronization of two fractional order chaotic systems. The proposed method uses the Radial Basis Function Neural Network(RBFNN) and proportional-integral(PI) parallel control to replace the traditional switching law of adaptive sliding mode control(ASMC). The advantages of the control method are as follows: firstly, the continuous nonlinear approximation capability of RBFNN ensures the system will not produce chattering since eliminating chattering is a key study point of ASMC; secondly, the superior self-learning ability of RBFNN makes it possible for the control systems to improve the overshoot and decrease the settling time; finally, the proposed method introduces the PI parallel control to overcome the disadvantage of local convergence of RBFNN. The asymptotic tracking of the states of the response system and the drive system is proved via the Lyapunov-based stability analysis. Considering the dimensions of the response system and drive system are not always same under normal conditions, the simulation results of two types are presented to validate the effectiveness of the method.

This paper proposes an enhanced robot skill learning system considering both motion generation and trajectory tracking. During robot learning demonstrations, dynamic movement primitives (DMPs) are used to model robotic motion. Each DMP consists of a set of dynamic systems that enhances the stability of the generated motion toward the goal. A Gaussian mixture model and Gaussian mixture regression are integrated to improve the learning performance of the DMP, such that more features of the skill can be extracted from multiple demonstrations. The motion generated from the learned model can be scaled in space and time. Besides, a neural-network-based controller is designed for the robot to track the trajectories generated from the motion model. In this controller, a radial basis function neural network is used to compensate for the effect caused by the dynamic environments. The experiments have been performed using a Baxter robot and the results have confirmed the validity of the proposed methods.

In this article, a sliding mode control (SMC) design based on a Gaussian radial basis function neural network (GRBFNN) is proposed for the synchronous reluctance motor (SynRM) system in electrical motorcycle applications. The conventional SMC assumes that the upper lumped boundaries of parameter variations and external disturbances are known, and the sign function is used. This causes high-frequency chattering and the high-gain phenomenon. In order to avoid these drawbacks, the proposed method utilizes the Lyapunov stability method and the steep descent rule to guarantee the convergence asymptotically, and reduce the magnitude of the chattering or avoid it completely. Finally, numerical simulations are shown to illustrate the good performance of our controller design.

This paper presents a robust adaptive sliding mode control strategy using radial basis function (RBF) neural network for a kind of value controlled asymmetric cylinder electro-hydraulic servo system of an ammunition transfer arm in the presence of uncertain nonlinearity and parameter uncertainty. On the premise of setting the expected trajectory, using RBF neural networks to approximate unknown parameters, by setting appropriate neural network parameters and adaptive terms, the change trend of position parameter was estimated. The stability of close loop system is verified by the Lyapounov theory. Compared with the simulation results of PID control method and expected trajectory, RBF Neural Net sliding mode control method has smaller system tracking error, faster response and better tracking performance.

Hydraulic system is widely used in industry where large actuation forces are needed such as electro-hydraulic positioning system, active suspension control, and so on. One of the main problem in hydraulic system is the nontriviality, for instance, the deadbands in valve operation, frictions of links, saturations, pressure dynamics, etc. both system unknown nonlinearities and external disturbances should be considered in controller design. For underwater hydraulic manipulator, because of the environment of sea and the large nonlinearities of the hydraulic manipulator system, both system unknown nonlinearities and external disturbances should be considered in controller design. A lot of control methods have been proposed to solve the problem properly, such as sliding mode control, adaptive control, fuzzy adaptive control, adaptive robust control, ect. However, due to limited knowledge about hydraulic manipulator system, it is hard to precisely estimate the nonlinearities for a possible higher performance. In this paper, a neural network adaptive robust control method is proposed to help control an underwater hydraulic manipulator. In order to achieve better performance using adaptive robust control strategy, unknown nonlinearities in the system are approximated by the outputs of radial basis function neural network (RBFNNs). A projection law has been used to bound the NN weights when they are tuning on-line to avoid possible adaptive divergence, and the tracking transient performance is guaranteed by robust control law. Simulation results illustrate that manipulator has a good tracking performance.

To meet the challenge of excellent tracking performance even under complicated nonlinear disturbance, a desired compensation neural network adaptive robust control (DCN-NARC) strategy is proposed for an industrial linear motor motion system. The proposed control scheme seriously concentrates on attenuating the effects of various model uncertainties and unknown nonlinearities, such as inertia load variation, force ripple, measurement noise, and external disturbances. Particularly, a three-layer radial basis function neural network (RBFNN) is employed to compensate the unknown nonlinearities. Online parameter adaptation is then utilized to reduce the effect of parametric uncertainties and to update the neural network weights. In addition, to facilitate the controller gain tuning process and attenuate the effect of measurement noise, the desired trajectory states are used in the adaptive model compensation following the desired compensation concept. The residual modeling error and unknown uncertainty are handled through certain robust control term. Theoretically, the proposed DCNNARC can achieve not only prescribed transient performance but also excellent tracking accuracy in general. Comparative experiments are conducted on an industrial linear motor system, and the results consistently demonstrate that the proposed scheme can achieve excellent tracking performance with remarkable disturbance rejection ability even under complicated disturbances. The proposed strategy also could be extended to other industrial control applications.