Design ideas and experiences gained during the development of an adaptive high precision track controller for ships are summarised. The controller is installed on a number of different ships, using a variety of position measurement systems. The design follows the spirit of the LQG paradigm combined with a model following feedforward strategy. The ship is kept to a manoeuvring trajectory through a combination of feedforward and LQG feedback control. The variances and weighting coefficients for the LQG controller are chosen systematically. The large number of design parameters is reduced by appropriate model scaling. In addition, a decomposition structure of the Kalman filter is exploited to reveal the important tuning parameters.

LQG approach for the high-precision track control of ships : Special section on marine control

The iris is considered as the biometric trait with the highest unique probability. The iris location is an important task for biometrics systems, affecting directly the results obtained in specific applications such as iris recognition, spoofing and contact lenses detection, among others. This work defines the iris location problem as the delimitation of the smallest squared window that encompasses the iris region. In order to build a benchmark for iris location we annotate (iris squared bounding boxes) four databases from different biometric applications and make them publicly available to the community. Besides these 4 annotated databases, we include 2 others from the literature. We perform experiments on these six databases, five obtained with near infra-red sensors and one with visible light sensor. We compare the classical and outstanding Daugman iris location approach with two window based detectors: 1) a sliding window detector based on features from Histogram of Oriented Gradients (HOG) and a linear Support Vector Machines (SVM) classifier; 2) a deep learning based detector fine-tuned from YOLO object detector. Experimental results showed that the deep learning based detector outperforms the other ones in terms of accuracy and runtime (GPUs version) and should be chosen whenever possible.

A Benchmark for Iris Location and a Deep Learning Detector Evaluation

This paper presents a course following control method for ships based on optimized backstepping (OB) technology. The backstepping technology is employed as the main control framework since the ship course can be modeled in the strict feedback form. Based on the actor-critic architecture and radial basis function (RBF) neural network (NN), the reinforcement learning (RL) strategy is introduced to avoid the difficulty in solving the traditional Hamilton-Jacobi-Bellman (HJB) equation directly. The actor NNs are used for carrying out the control law, while the critic NNs aim at evaluating the tracking performance. An auxiliary design system and Gaussian error function are employed to handle the practical problem of input saturation. The stability of the closed-loop system can be guaranteed via Lyapunov theory. Finally, simulation examples and comparison are provided to demonstrate and verify the superior performance and advantages on course following and energy saving of the control scheme proposed in this paper.

/pdf/optimized-backstepping-design-for-ship-course-following-2hm9owvayl.pdf

Optimized Backstepping Design for Ship Course Following Control Based on Actor-Critic Architecture With Input Saturation

Safe navigation in a coastal environment of multiple surface vehicles under uncertainties: A combined use of potential field constructions and NMPC

A model predictive controller with disturbance observer is presented to force an automated unmanned surface vessel (USV) to follow a reference path with environment disturbance. A full-scale trials by fully instrumented USV is carried out to identify the dynamic model. The Serret-Frenet frame is used to define the tracking error, therefore the position tracking errors can be stabilized by stabilizing the state of the control model to zero. And the constrained control input of the considered system is solved by minimizing performance based on model predictive control (MPC). Simulation and experiments results are presented to validate the effectiveness of the proposed method.

Model predictive controller design with disturbance observer for path following of unmanned surface vessel

Iris recognition is one of common biometric identification technologies, and an important part of iris recognition is iris location, whose precision directly affects the accuracy of iris recognition. Fitting the inner and outer edge of iris as an approximate circle is a common iris location method, which thereby can lead to fitting errors between the location result and the actual edge. In order to solve this problem, the paper proposes an iris location algorithm based on regional property and iterative searching. The pupil area is extracted using the regional attribute of the iris image, and the iris inner edge is fitted exactly by iterating, comparing and sorting the pupil edge points; then the outer edge location is completed in an iterative searching method on the basis of the extracted pupil centre and radius. The results of the experiment show that in this algorithm, more accurate iris inner edge location can be achieved and a range of comprehensive advantages of faster location speed and higher location accuracy are.

Zhilin Liu

Papers

Model predictive controller design with disturbance observer for path following of unmanned surface vessel

Iris location based on regional property and iterative searching