Search

As an increasing number of unmanned aerial vehicles (UAVs) have been widely applied in many aspects, controllers with higher performance are preferred. In this paper, a new varying-parameter adaptive multi-layer neural dynamic based controller (termed as VP-AMND controller) design method is proposed and applied to controllers of multi-rotor UAVs. First, a varying-parameter convergent neural dynamic (VP-CND) based controller is proposed and its convergence and robustness are theoretically proven. Second, by incorporating the adaptive control method into the VP-CND controller, the VP-AMND controller design method is proposed, of which the global stability, fast convergence speed and strong robustness can be guaranteed. Different from traditional triple zeroing dynamic (TZD) and VP-CND controllers, the proposed VP-AMND controller with self-tuning rates can estimate the unknown disturbances and enhance the stability of the system in the face of uncertainty. Third, computer simulation results verify that the multi-rotor UAVs with VP-AMND controllers can track time-varying trajectories quickly and solve the parameter uncertainty and disturbances problems effectively.

Two common methods exist for solving indoor autonomous navigation and obstacle-avoidance problems using monocular vision: the traditional simultaneous localization and mapping (SLAM) method, which requires complex hardware, heavy calculations, and is prone to errors in low texture or dynamic environments; and deep-learning algorithms, which use the fully connected layer for classification or regression, resulting in more model parameters and easy over-fitting. Among the latter ones, the most advanced indoor navigation algorithm divides a single image frame into multiple parts for prediction, resulting in doubled reasoning time. To solve these problems, we propose a multi-task deep network based on feature map region division for monocular indoor autonomous navigation. We divide the feature map instead of the original image to avoid repeated information processing. To reduce model parameters, we use convolution instead of the fully connected layer to predict the navigable probability of the left, middle, and right parts. We propose that the linear velocity is determined by combining three prediction probabilities to reduce collision risk. Experimental evaluation shows that the proposed method is nine times smaller than the previous state-of-the-art methods; further, its processing speed and navigation capability increase more than five and 1.6 times, respectively.

In this paper, a feedback control law based on two time‐scale assignment and state extension is implemented for the nonlinear non‐minimum phase dynamics of three degree‐of‐freedom autonomo...

In this paper, we focus on the transition control of a ducted fan vertical take-off and landing (VTOL) unmanned aerial vehicle (UAV). To achieve a steady transition from hover to high-speed flight, a neural-networks-based controller is proposed to learn the system dynamics and compensate for the tracking error between the aircraft dynamics and the desired dynamic performance. In prior, we derive the nonlinear system model of the aircraft full-envelope dynamics. Then, we propose a novel neural-networks-based control scheme and apply it on the underactuated aircraft system. Key features of the proposed controller consist of projection operator, state predictor and dynamic-formed adaptive input. It is proved and guaranteed that the tracking errors of both state predictor and neural-networks weights are upper bounded during the whole neural-networks learning procedure. The very adaptive input is formed into a dynamic structure that helps achieve a reliable fast convergence performance of the proposed controller, especially in high-frequency disturbance conditions. Consequently, the closed-loop system of the aircraft is able to track a certain trajectory with desired dynamic performance. Satisfactory results are obtained from both simulations and practical flight test in accomplishing the designed flight course.

In this paper, a novel disturbance-observer-based approximate dynamic inversion (ADI) approach is developed for pure-feedback nonaffine-in-control nonlinear systems (PFNNSs) in the presence of both high-order mismatched disturbances and actuator saturation. Finite-time disturbance observers (FTDOs) are utilized to estimate the disturbances and their derivatives. Then, we rebuild the system with the outputs of FTDOs. Thereafter, ADI is employed to derive the desired virtual and actual control of the nominal system, where no disturbances and saturation are presented. Furthermore, an augmented intermediate subsystem is constructed for the reduced slow subsystem to compensate for the difference between inputs with and without saturation by approximating its inversion. The stability of the closed-loop system is studied using Tikhonov's theorem. The proposed method is applied to a numerical example and a one-link robotic system with a brush DC motor. The simulation results demonstrate the validity of the presented approach.

Relative pose estimation refers to estimate the relative attitude and translation between multiple platforms. For mobile platforms, tracking the relative pose with pairwise range is challenging for highly nonlinear associations between measurement and state. This article proposes a promising framework using pairwise range to estimate the relative pose parameterized with Lie algebra. It is compatible with the existing Gauss-Newton method and the Levenberg–Marquardt method. We analyze the existence of the optimal solution based on the rank of the Hessian matrix, which turns into a discussion of sensors placement. The associated unconstrained Cramer-Rao Lower Bound with fewer variables is presented. To track moving platforms, we derived a novel and accurate relative kinematics without angular accelerations. An extended Kalman filter incorporating the measurement of an IMU is designed to generate smooth poses. A simplified version of the optimizer with less dimension is introduced to the application of aerobridge, which is also compatible with other multilink devices. Simulations verify the proposed algorithm and the comparisons with the existing popular methods prove its novelty.

In this paper, we focus on the hover/high-speed-cruising transition flight control of a ducted fan unmanned aerial vehicle (UAV), in which saturation on control surfaces is taken into consideration. To achieve full-envelope flight control, we propose an adaptive cascaded structure on the velocity tracking control for the nonlinear aircraft system. By introducing adaptive neural network, the closed-loop system tracks a prescribed reference system with all the nonlinear tracking errors well canceled. In the meanwhile, to tackle the control surface saturation, we propose a control effectiveness enhancement (CEE) algorithm in attaining sufficient control moment to the ducted fan vehicle. Under the proposed cascaded adaptive control scheme in conjunction with the CEE algorithm, the state tracking performance is proved to be bounded when encountering surface saturation. Finally, comparison experiments in terms of the forward and backward flight transition are conducted on the ducted fan UAV. The results are satisfactory and convincing in exhibiting the flight transition course that demonstrates the effectiveness of the proposed method.

To handle the problems of point cloud data quantity explosion and urgent need for real-time access to massive point cloud data in collecting and storing phase, this paper proposes an index technology with internal and external storage interactions, which is based on extended 3D grid in memory layer and B-tree in external storage layer. The experiment shows that it is efficient to retrieve massive point cloud data by using this double-layer index structure.

The autopilot of aerobridge has attracted considerable attention recently since human error and high labor costs. To estimate the pose of an aerobridge, this paper proposes a navigation algorithm with the aid of pairwise distances and inertial measurements. We first locate the joints of the aerobridge with the distance measurement. Then the pose of the aerobridge is calculated using its kinematic in closed-form and the associated covariance is derived. A dual quaternion based multiplicative extended Kalman filter is designed to fuse inertial measurements with the estimated pose, which produces a smooth pose. To verify the precision of this algorithm, a simulation is constructed based on Robotics Toolbox and the experiment results are positive.

In this paper, we generalize the Extended Kalman Filter algorithm on Lie Groups for the attitude propagation and update of the unmanned helicopters. The data noises obtained by sensors are assumed to be independent as Gaussian distributions, and the posterior distribution of the estimated state is also Gaussian distribution. The attitude of the drone can be obtained from the accelerometers, the gyroscopes, and the magnetometers. These three sensors need to be fused to obtain the optimal estimated value. Among them, the method of multiple sub-samples is used to compensate for the error generated by the long sampling period of the gyroscope, and we use a Chi-square detection method to determine whether the observation information is valid during the updating process.

At present, the main methods of solving the monocular depth estimation for indoor drones are the simultaneous localization and mapping (SLAM) algorithm and the deep learning algorithm. SLAM requires the construction of a depth map of the unknown environment, which is slow to calculate and generally requires expensive sensors, whereas current deep learning algorithms are mostly based on binary classification or regression. The output of the binary classification model gives the decision algorithm relatively rough control over the unmanned aerial vehicle. The regression model solves the problem of the binary classification, but it carries out the same processing for long and short distances, resulting in a decline in short-range prediction performance. In order to solve the above problems, according to the characteristics of the strong order correlation of the distance value, we propose a non-uniform spacing-increasing discretization-based ordinal regression algorithm (NSIDORA) to solve the monocular depth estimation for indoor drone tasks. According to the security requirements of this task, the distance label of the data set is discretized into three major areas&mdash, the dangerous area, decision area, and safety area&mdash, and the decision area is discretized based on spacing-increasing discretization. Considering the inconsistency of ordinal regression, a new distance decoder is produced. Experimental evaluation shows that the root-mean-square error (RMSE) of NSIDORA in the decision area is 33.5% lower than that of non-uniform discretization (NUD)-based ordinal regression methods. Although it is higher overall than that of the state-of-the-art two-stream regression algorithm, the RMSE of the NSIDORA in the top 10 categories of the decision area is 21.8% lower than that of the two-stream regression algorithm. The inference speed of NSIDORA is 3.4 times faster than that of two-stream ordinal regression. Furthermore, the effectiveness of the decoder has been proved through ablation experiments.

In this paper, we address the hover-to-cruise transition control problem for high-speed level flight of a ducted fan UAV. Firstly, the nonlinear aircraft dynamic system is derived and relevant key aerodynamic characteristics are presented. Then, we divide the dynamic system into known parts and uncertain parts. In order to track a reference system formed by the known parts, we propose a neural-network-based control strategy to learn the error dynamics and cancel the tracking error with proved stability. Finally, simulations and experiments are conducted to perform the transition process at fixed altitude and constant horizontal acceleration, with satisfactory results accomplishing the designed flight course.

This paper presents a new recursive identification method which can efficiently estimate time-varying parameters in discrete time systems and has significant advantages over standard recursive leas...

This paper addresses the robust and accurate trajectory tracking problem for unmanned helicopters in the presence of model uncertainties and external disturbances. First, the helicopter's model is simplified to a six-degrees-of-freedom rigid body augmented with a simplified rotor dynamic model, with the model uncertainties and the external disturbances being treated as lumped unknown disturbances. Second, a nonlinear disturbance observer is designed to estimate this lumped disturbance. Then, a backstepping controller with disturbance compensation is designed to ensure robust and highly trajectory tracking. After that, the theoretical analysis of the efficiency of the designed disturbance observer-based backstepping controller (Backstepping+DO) is shown by the Lyapunov theory. Finally, simulation results and conclusions are presented and discussed.

In order to simulate the in-orbit movement of formation satellites, the unmanned helicopter is utilized to set up the microgravity environment relative navigation ground simulation platform. The platform can provide the movement of six degrees of freedom and can simulate the formation satellite flying in three-dimensional space. The test platform includes host computer system, communication system, UAVs formation control system, satellite attitude simulation system, and so on. Sensors are equipped with laser radar and visible light cameras. Formation control and navigation of multi-body can be achieved under complex network environment. The platform can be used for the test of satellite autonomous formation simulation, joint fleet navigation, and so on. The algorithm of multi-body formation, multi-source and multi-target tracking based on sensor network, multi-body navigation and remote control can be verified on this platform.

We address the problem of environmental contour line tracking for a class of autonomous vehicles. A reference velocity is designed for the autonomous vehicles to do contour line tracking. Based on Lashall invariance principle, an ideal controller is designed for the vehicle with ideal model and ideal information about the environmental concentration function to track the desired contour line. For the vehicle with possibly modeling uncertainty, we combine a neural controller containing a wavelet neural network (WNN) identifier with a robust control to construct a robust adaptive WNN control for the vehicle to track the desired environmental contour line. Then we give theoretical proof of the efficiency of the designed robust adaptive WNN control. Simulation results and conclusion are presented and discussed.

This paper addresses the robust trajectory tracking problem for a robot manipulator in the presence of uncertainties and disturbances. First, a neural network-based sliding mode adaptive control (NNSMAC), which is a combination of sliding mode technique, neural network (NN) approximation and adaptive technique, is designed to ensure trajectory tracking by the robot manipulator. It is shown using the Lyapunov theory that the tracking error asymptotically converge to zero. However, the assumption on the availability of the robot manipulator dynamics is not always practical. So, an NN-based adaptive observer is designed to estimate the velocities of the links. Next, based on the observer, a neural network-based sliding mode adaptive output feedback control (NNSMAOFC) is designed. Then it is shown by the Lyapunov theory that the trajectory tracking errors, the observer estimation errors asymptotically converge to zero. The effectiveness of the designed NNSMAC, the NN-based adaptive observer and the NNSMAOFC is illustrated by simulations.

In order to realize the localization demand of mobile robot, the motion model and observation model of mobile robot were analyzed, and the mobile robot real-time localization algorithm of given map environment based on extended Kalman filtering was described. Besides, the algorithm was simulated. The landmark model of map was built in the process of simulation, and the landmark that was observed under control action was used to correct the location of mobile robot, by which the positioning of robot was realized. The result of simulation shows that the positional accuracy of this algorithm is very high. In addition, the construction method of real-time map of mobile robot was described on extended Kalman filtering under unknown environment based on this. Experimental result shows that this algorithm can identify the location of landmark accurately so as to give reference for the real-time positioning and mapping construction of mobile robot.

The core of the abstraction-refinement verification is the analysis of the counterexample and refinement of the abstract model, which are the most expensive parts of the verification procedure. In this paper, we present a method that aims at reducing the complexity as much as possible. Our method combines the qualitative analysis with the reachability analysis. Qualitative analysis checks the feasibility of pair-wise transitions without the explicit numerical reachability computation. While reachability analysis employs bidirectional analysis strategy to minimize the region to be refined. Experiment shows that the integration of these strategies and techniques make the verification more efficient.

In vision-based autonomous landing system of UAV, the efficiency of target detecting and tracking will directly affect the control system. The improved algorithm of SURF(Speed Up Robust Features) will resolve the problem which is the inefficiency of the SURF algorithm in the autonomous landing system. The improved algorithm is composed of three steps: first, detect the region of the target using the Camshift; second, detect the feature points in the region of the above acquired using the SURF algorithm; third, do the matching between the template target and the region of target in frame. The results of experiment and theoretical analysis testify the efficiency of the algorithm.