Search

This paper deals with longitudinal motion control of electric vehicles (EVs) driven by in-wheel-motors (IWMs). It shows that the IWM-EV is fundamentally a multi-agent system with physical interaction. Three ways to model the IWM-EV are proposed, and each is applicable to certain control objectives. Firstly, a nonlinear model with hierarchical structure is established, and it can be used for passivity-based motion control. Secondly, a linearized model with rank-1 interconnection matrix is presented for stability analysis. Thirdly, a time-varying state-space model is proposed for optimal control using linear quadratic regulator (LQR). The proposed modellings contribute the new understanding of IWM-EV dynamics from the view point of multi-agent-system theory. By choosing the suitable control theory for each model, the complexity level of system design is maintained constant, no matter what the number of IWMs installed to the vehicle body. The effectiveness of three models and their design approaches are discussed by several examples with Matlab/Carsim co-simulator.

This paper proposes a distributed energy management system-based algorithm to solve the optimal schedule of multiple combined heat and power systems in which exponential efficiencies of fuel cells are considered. To deal with the variation in fuel cell efficiencies, the proposed algorithm utilizes the alternating direction method of multipliers, consensus theory, and quadratic programming to solve the optimization problem with constant efficiencies. Then, the efficiency matching is checked iteratively to verify the obtained result. The optimization algorithm is constructed in a whole-day distributed form, such that all time slot variables can be solved simultaneously in a distributed way. With the help of this simultaneous solution method, operation of hot tanks and combined heat and power systems can be scheduled globally. The proposed algorithm is tested successfully with a test system having 4 combined heat and power systems, showing fast convergence in numerous simulations.

This paper proposes a novel optimization model and a novel approach to derive new demand response (DR) and real-time pricing schemes for smart grid in which renewable energy and power losses are taken into account. In our proposed optimization model, a time-varying load constraint is introduced to better capture the consumption variation of customers and hence gives our approach an adaptive feature as well as facilitates DR. Then our approach enables all generation and demand units to actively collaborate in a distributed manner to obtain the optimal electric price and their optimal power updates in real-time while achieving their best profits. To do so, the total welfare in the grid is maximized and the optimization problem is analytically solved using the alternating direction method of multipliers and consensus theory for multi-agent systems. Moreover, the power balance constraint is guaranteed in every iteration of the proposed algorithm. Next, the effects of renewable energy to conventional generation, consumer consumption, and electric price are theoretically revealed which show the essential role of renewable energy for peak load shifting. Finally, simulations on the IEEE 39-bus system are introduced to illustrate the effectiveness of the proposed approach.

In this work, we address the problem of robot pursuit based on a real-time object recognition system with 3D depth sensors. Compared with traditional RGBD data based recognition approaches, we propose a novel global online descriptor designed for object recognition from solely depth data. Proposed descriptor, which we name as Differential Histogram of Normal Vectors (DHONV), is designed to extract the geometric characteristics of the captured 3D surfaces of the objects presented in depth images. In order to obtain a brief description of the visible 3D surfaces of each object, we quantize the differential angles of the surface’s normal vectors into a 1D histogram. The object recognition experiments on a self-collected dataset and a benchmark RGB-D object dataset show that our proposed descriptor outperforms other depth data based descriptors. Moreover, we conducted real-time experiments with RoboCars. Our experiments with RoboCars validate our proposed method capability to perform a real-time recognition and pursuit tasks within indoor environment based solely on depth data.

Depth data based object recognition has recently emerged as a challenging research topic. In this work, we develop a novel approach to perform detection and recognition of occluded 3D objects. We propose a hierarchical segmentation algorithm in order to obtain the homogeneous sub-regions contained in each depth frame which in turn facilitates the recognition under severe occlusion conditions. Our proposal consists of three steps: the first step is to build a tree structure contains all key sub-surfaces visible in the depth frame with their intra-hierarchical relations. Thereafter, we draw a classification prediction for all nodes based on a combination of convolution and recursive neural networks. Finally, we employ the hierarchy scheme to refine the classification results. Our proposal obtained competitive results and proved to be invariant to objects scale, rotation, and viewpoint variations.

Object recognition from depth sensors has recently emerged as a renowned and challenging research topic. The current systems often require large amounts of time to train the models and to classify new data. In this work, we present a novel fast approach for object recognition from 3D data acquired from depth sensors such as Structure or Kinect sensors. We first extract simple but effective frame-level features from the raw depth data and build a recognition system based on extreme learning machine with a local receptive field. We test the presented method on two datasets: A self-collected dataset and the benchmark RGB-D object dataset. Experiments on both datasets show the effectiveness of our presented approach compared with the state-of-the-art methods. Fast computational time and high recognition accuracy make the presented method readily applicable for online recognition applications.

Velocity field control (VFC) is an alternative approach for motion control of robotic systems. It has advantages over the trajectory tacking problem when the timing in the desired task is of less importance in the application of interest. Recently, this control strategy has been emerged as a promising control method in robot-aided rehabilitation. Therefore, this paper addresses the problem of VFC for robotic exoskeletons with dynamic uncertainties. Most existing VFC methods require some knowledge of the dynamic model of the robot, which is usually difficult in practice to precisely identify. Consequently, this paper design an adaptive neural network VFC method to compensate for the dynamic uncertainties. The controller gives a priori bounded control command in order to take into account the saturation of actuators. The controller performance is validated through simulation and experimental studies on a two-degree-of-freedom lower-limb robotic exoskeleton.

Current consumer depth sensors produce depth maps that are often noisy and lack sufficient detail. Enhancing the quality of the 3D depth data obtained from compact depth Kinect-like sensors is an increasingly popular research area. Although depth data is known to carry a signal-dependent noise, the state-of-the-art denoising methods tend to employ denoising techniques which are independent of the depth signal itself. In this paper, we present a novel adaptive denoising filter to enhance object recognition from 3D depth data. We evaluate the performance of our proposed denoising filter against other state-of-the-art filters based on the enhancement of object recognition accuracy achieved after denoising the raw data with each filter. In order to perform object recognition from depth data, we make use of Differential Histogram of Normal Vectors (DHONV) features along with a linear SVM. Experiments show that our proposed filter outperformed the state-of-the-art de-noising methods.

This paper addresses the assist-as-needed (AAN) control problem for robotic orthoses. The objective is to design a stable AAN controller with an adjustable assistance level. The controller aims to follow a desired trajectory while allowing an adjustable tracking error with low control effort to provide a freedom zone for the user. By ensuring the stability of the system and providing the freedom zone, the controller combines the advantages of both model-based and non-model-based AAN controllers existing in the literature. Furthermore, the controller provides a priori bounded control command, and includes an adaptive neural network term to compensate for the uncertainties of dynamic model of the system, mainly when a precise tracking is of interest. The stability of the closed-loop system is well analysed based on the Lyapunov method. The effectiveness of the proposed control scheme is validated through experiments using a lower extremity robotic exoskeleton.

This paper proposes a distributed control approach for optimal power dispatch towards environment-friendly power grids. To do so, an optimization problem called environmental economic dispatch, which is subjected to physical constraints of the considered power system, is solved. Consequently, we propose an approach based on the Alternating Direction Method of Multipliers (ADMM) to obtain the globally optimal solution of that optimization problem in a distributed manner under a fast convergence. As a result, each generation unit can derive by itself an optimal generated power that minimizes with compromising both fuel and emission costs while satisfying both global and local physical constraints caused by the whole system and by its own. The most distinguished feature of this approach is that the power balance constraint is always guaranteed during the execution of the algorithm. Finally, the performance of our proposed approach is illustrated through the simulation to a realistic power system and the comparison with another method.

This paper aims at proposing a distributed control approach for optimal power dispatch towards environment-friendly power grids. In order to do so, both electric generation and pollutant emission costs are properly incorporated into a unique objective function of an optimization problem subjected to physical constraints of the considered power system. Consequently, we propose an approach based on the Alternating Direction Method of Multipliers (ADMM) to obtain the globally optimal solution of that optimization problem in a distributed manner under a fast convergence. As a result, each generation unit in a power system can derive by itself an optimal generated power that minimizes with compromising both fuel and emission costs while satisfying both global and local physical constraints caused by the whole system and by its own. The most distinguished feature of this approach is that the power balance constraint is always guaranteed during the execution of the algorithm. Finally, the performance of our proposed approach is illustrated through the simulation to a realistic power system and the comparison with another method.

Due to ill-defined and poorly understood dynamics of flapping wing flying robots, acquiring high-fidelity aero dynamical and control models for such aerial robots are challenging and often infeasible. In this work, we propose using a non-linear, model-free approach that eliminates errors arising from the mis-modeling or under-modeling of flapping dynamics. These procedures require no prior knowledge or model of any of the dynamical aspects of the flying robot and assume zero statistical knowledge about noise affecting sensor measurements. Intelligent PID (iPID) controllers are used to provide an abstracted treatment of all unknowns and uncertainties in the robot by encapsulating them together into a single term. The problem then devolves into the estimation of this term by either direct or indirect means which is more practical than obtaining high-fidelity models. For indirect estimation, numerical differentiation is required and we propose using a Haar-wavelet differentiator that is capable of smoothing and differentiating signals irrespective of the multivariate noise corrupting the signal. The direct estimation method uses an algebraic and non-asymptotic method. Pitch control and altitude control experiments are successfully performed on a real, bird-like flapping wing flying robot using all different approaches; results obtained reveal that indirect estimation using the Haar-wavelet differentiator provides slightly better performance.

Rigid planar space robot is represented as Caplygin system and its base attitude can be stabilized by means of geometric phase. However, in cases when flexible arms are used for control of base attitude, inherent resonance modes are excited and instability is resulted. In this study we propose a stabilized controller is synthesized as adaptive tracking control system convined with sliding mode control to track to the arm trajectories derived from the geometric phase of the Caplygin system. Usefulness of the proposed controller is demonstrated by numerical simulations and laboratory experiments.

Electric Vehicle (EV) which has a large amount of storage battery can contribute to make Energy Management System (EMS) more effective. While an EV is parking at home or at company, the battery of an EV can be used as an energy storage for smart grids. In this paper, we propose a model predictive control of EV storage battery in Home Energy Management System (HEMS) based on an auto-regressive model and Particle Swarm Optimization (PSO). We also study the effectiveness of Photovoltaics (PV) equipped with EVs. The proposed method controls the states of EV storage battery to minimize the electricity cost of HEMS and evaluates the electric company's profit.

This paper proposes a novel approach to design distributed optimal controllers for the power coordination problem in Wind Power Plants (WPPs). Based on LQR method, we present a systematic procedure to design a fully distributed optimal stabilizing controller for the linearized model of Wind Turbines (WTs) in a WPP. Next, the reduced-order distributed optimal controller is derived by elaborating more on the selection of weighting matrices in the LQR performance cost. Furthermore, the effect of disturbances to the system performance is evaluated through the H ∞ norm computation. Finally, the simulation results for a linearized model of WPP are shown to illustrate the effectiveness of the proposed approach.

In this paper we consider the problem of plant parameter estimation for a flapping wing flying robot. These parameters are the coefficients of polynomials appearing in transfer functions that govern some robot dynamics e.g pitching, and their values are completely unknown and even susceptible to variation over time. First, we analyse the pitching dynamics of the SlowHawk2 flapping wing flying robot under study and derive the linearized state-space model. Our analysis shows that in addition to the elevator deflection inducing pitching motion on the flying robot, extraneous pitching is also caused by the flapping wings. Thus, our derivations yield a multiple-input single-output (MISO) system consisting of two uncoupled transfer functions where both the elevator deflection and the wing-flapping action are the inputs that determine the pitching. Subsequently, an online parameter estimator is designed for the MISO system in order to estimate values of coefficients in the transfer functions. We use the recursive least squares method with the so-called forgetting factor. How the parameter estimator is designed using only the known input signals and measured output signals is described in detail. Using real sensor data obtained from real test flights, the designed estimator shows that the values of the unknown parameters can be accurately and robustly determined.

Recognizing the human gait is an edge research in robotics and rehabilitation. It has been popular to recognize the human gait from kinematic data. However, the recognition from muscle activities, the input of the movement, has not been widely approached. In this paper, we propose a framework to recognize the human walking and running movements by investigating muscle activities through electromyography (EMG). The framework is a Hidden Markov Model (HMM) topology utilizing Locally Linear Embedding (LLE) technique to extract feature vectors. We show that: (1) the high-dimensional EMG data can be embedded into a lower-dimensional space by using a manifold learning algorithm (LLE), primitive components which give meaningful representation of the EMG can be extracted, and (2) our proposed HMM topology whose input are the extracted vectors from EMG can recognize the gait movement at an accuracy rate of over 80%.

This paper proposes a control strategy for descriptor non-polynomial systems using parallel asynchronous particle swarm optimization (PAPSO). Basic idea of our approach is to extend the existing approaches for estimating the domain of attraction (DOA) of the non-polynomial systems to synthesis of stabilizing control. To do this we derive the stability conditions for estimating the DOA with input magnitude constraints. From these conditions stabilizing controller is obtained by PAPSO. The proposed strategy can be easily exploited to search for both the stabilizing controller and optimal estimates of DOA. Usefulness and validity are demonstrated by numerical simulations.

An understanding of muscle activities can reveal the mechanism of producing locomotion, helping to improve the performance of assistive robots in supporting the users' movement. This paper proposes a framework to recognize the human walking movement by investigating electromyography (EMG), which has not been widely approached yet. The framework is a combination of a Hidden Markov Model (HMM) and Locally Linear Embedding (LLE) technique. First, we show that using LLE algorithm we could reduce the dimensionality of the high-dimensional EMG dataset. The extracted primitive components gave a meaningful representation of the EMG. Second, we demonstrate that the HMMs trained by these components could recognize the movement intention at a high rate of accuracy.

A trajectory specification method for a flapping-wing flying robot exhibiting a novel dynamical model than other types of micro aerial vehicles such as quadrotors and airplanes is presented. The desired trajectory is first defined as a vector-valued, differentiable parametric curve of the form f : RR → RR. The velocity vector at any point of the trajectory is the same as the unit tangent vector obtained using parametric differentiation. The acquired unit tangent vector is subsequently expressed in quaternion form and nominal values for the pitch and yaw angles at every point of the trajectory are determined. Finally, the feed-forward trajectory is specified as a function of the input control variables (δr, δe) of the flapping wing flying robot. At the same time, the feasibility of a specified trajectory is verified so that there are no violations of intrinsic constraints. Both two and three dimensional flight trajectories are simulated for exemplification. Simulation results are compared to real sensor data, thus verifying the procedure.

This paper aims to describe a framework implementing Multi-Class Support Vector Machine (MCSVM)-based motion intention recognition. To this end, we primarily constructed a wearable exoskeleton robot of lower body (TTI-Exo) which is employed as the experimentation platform to test the proposed method of motion intention recognition based on MCSVM and the assist effectiveness as well. Experiments of stand-to-sit and sit-to-stand movements were carried out to test the MCSVM method and TTI-Exo's motion assist. Having disclosed prototype development, experimental results are presented. We verified that our proposed method based on MCSVM obtained a better recognition accuracy than a conventional method based on threshold values. Muscle activities when subjects wearing TTI-Exo were much smaller than when subjects not wearing the exoskeleton, thus implying the assist efficacy of our power assist system.

This paper describes an active jiggle control of hung structure using a control moment gyro (CMG) which has two gimbals for one flywheel. The double gimbal CMG enables us to simultaneously control the vibrations for two different dirctions by only one flywheel. Due to the nonlinearlity and interference, it is however difficult to construct the control system for the double gimbal CMG. First, we propose a simple control system design method for the double gimbal CMG based on H2 optimal control theory. In order to improve the gimbal stability, which is necessary for avoiding the interference, we then propose an another design method based on H2/H∞ control using LMI optimization. The performance of all designed control system are evaluated by simulations and experiments.

This paper is written to report our research group's recent activities that are concerning quadruped locomotion control. To this end, we primarily constructed an electrically-actuated quadruped robot which is employed as an experimentation platform to test the locomotion control algorithm. An overall motion control scheme is introduced to reveal the main principles for achieving fast and agile locomotion scenarios. Having disclosed prototype development and real-time control procedures, trot-running locomotion experimental results are presented. In these experiments, the robot exhibited successful trot-running cycles in a repetitive, dynamically-equilibrated, agile, and compliant manner; demonstrating that the control algorithm has potentials to be utilized in fast locomotion tasks.

In this paper, we consider an analysis and design method for the robust controller of magnetic levitation systems with parameter perturbations. First, an expression of the uncertain plant is given by taking account of the physical parameter variations directly. Then, based on this result, an analysis/design method for robust controller is given in the framework of μ, where the unmodeled dynamics and the sensitivity performance are also taken into consideration. Second, the effectiveness of the proposed method is evaluated by experiments.

In this study we give a feedback control method for steering support system. In this method we consider the stability of the path tracking control system under the limitation of steering angle, so that drivers can drive more safety and comfortable. In order to do this, we have derived linearized system from the nonholonomic kinematic system by the flatness consideration and linear control law which stabilizes the path tracking control system. Furthermore, feedback gains are tuned to satisfy the closed loop stability and limitation of the steering angle. Usefulness of the proposed control method has been demonstrated by experiments using 1/10 scaled robot car.

This paper is aimed at presenting a framework that consists of a pattern generator and a controller, which are combined together to realize continuous and dynamically equilibrated running motion on a 4-link 3-DoF one-legged robot with no passively compliant elements. Initially, we make use of a pattern generator to synthesize dynamically-consistent running trajectories in which the rotational inertia and the associated angular momentum term are characterized. As for the controller, ground reaction force constraints are imposed to the system indirectly. For this purpose, joint torque values that are corresponding to horizontal and vertical force errors are computed. Subsequently, they are inserted to an admittance filter block to obtain the associated joint displacements. These joint displacements are then fed-back to local servo controllers to implement indirect force feedback control in an actively compliant manner. Additionally, friction compensation and foot orientation controller blocks are added to enhance the system performance. In order to validate the method, running experiments are conducted on the actual one-legged robot. As the result, we satisfactorily obtained continuous, dynamically equilibrated and repetitive running cycles.

In parallel mechanisms with redundant actuators, actuator forces are underspecified because of redundancy. In this paper, the method to minimize a maximum actuator force was proposed. At first, parallel mechanism with two redundant actuators and Stewart platform model using the linear shaft motor were developed. This redundant parallel mechanism was consists of Stewart platform and additional 2 redundant actuators. Inverse dynamics of each mechanisms were formulated. Secondly, effectiveness of dynamic control of Stewart platform was experimentally confirmed. Finally, the method to minimize a maximum actuator force using Linear Matrix Inequality (LMI) was proposed. And the effectiveness of the proposed method was confirmed by simulation.

In this paper, we propose a data-based PID controller design method using unfalsified control technique with Support Vector Machine (SVM) and gap metric. SVM and gap metric are utilized for adjusting the weighting function of extended L2 gain criterion. The effectiveness of the proposed method is evaluated by experiments on a magnetic levitation system and an active magnetic bearing system.

This paper gives a design example of multipurpose controllers for two-mass-spring systems based on BMI optimization technique. The obtained controller achieves (a) tracking with zero steady-state error, (b) pole placement in a prescribed region, and (c) optimal H/sub /spl infin// performance, simultaneously in the presence of physical parameter uncertainties. In addition, its effectiveness is demonstrated via simulations and experiments.