Search

Reinforcement learning (RL) is suitable for the design of unmanned surface vessels (USV) path-following controllers for its model-free and unsupervised learning. However, the USV state transition is not strictly consistent with Markov properties and the state-space, action-space, reward function strongly affect the performance of the RL-based controller. According to the dynamics and kinematic characteristics of USV, we design a new state-space to reduce the influence of large inertia and state hysteresis on the RL agent training. A comprehensive reward function is proposed to avoid the RL-based controller falling into local optimum according to the task decomposing. A dynamic threshold is used in the reward function to accelerate the training speed while ensuring tracking accuracy. Finally, the effectiveness of the proposed RL-based controller is evaluated by means of simulation and actual USV path following. • According to the USV dynamics and kinematic characteristics, we design a new state-space to make state transitions Markovian and to reduce the influence of large inertia and state hysteresis on training. • By decomposing the path following task into primary task and secondary tasks, a comprehensive reward function is proposed to make the RL-based agents converge quickly and avoid falling into local optimum. • A dynamic threshold is used in the reward function to avoid the sparse rewards and accelerate the training speed while ensuring tracking accuracy. [ABSTRACT FROM AUTHOR]

To quickly evaluate soybean quality, we proposed a deep learning-based method for online classification of soybean seeds. Firstly, images of soybean seeds with uneven illumination were segmented based on the multi-scale Retinex with color restoration (MSRCR). Then, a convolutional neural network (CNN) was constructed to achieve soybean seed four-classification with appropriate parameters. The F-score of the normal, damaged, abnormal, and non-classifiable soybeans reached about 95.97%, 97.41%, 97.25%, and 96.14%, respectively. Finally, the method was successfully applied in NVIDIA Jetson TX2 with an accuracy of 95.63% and an average classification time of 4.92 ms for a soybean seed, which can meet the requirement of online soybean quality assessment. [ABSTRACT FROM AUTHOR]

In order to enhance the large curvature path following performance of unmanned surface vehicle (USV), an improved Stanley guidance law (ISGL) is proposed. To the best of our knowledge, we are the first to introduce the Stanley guidance law (SGL) to guidance subsystem of USV, and theoretically prove that it can be unified into one form with the line-of-sight (LOS). It is essentially a kind of LOS method with time-varying lookahead distance in the path-tangential reference frame. For the sake of accurately following the path with large curvature, a compensation angle (k m / (1 + | y e |)) Δ α m is integrated to improve the original SGL, which can evaluate the path curvature in advance and compensate the desired heading angle. The aim is to improve the heading angle error and make USV turn in advance. In order to verify path following performance of our algorithm, theoretical simulation and lake experiments are launched to compare to extended line-of-sight (ELOS), integral line-of-sight (ILOS) and original SGL algorithms. All the results show that our ISGL algorithm has better path following performance than others. • We have proved theoretically for the first time that Stanley guidance law is equivalent to be LOS with time-varying look-ahead distance in the path-tangential reference frame. • Stanley guidance law is improved and introduced to the guidance subsystem of USV to advance the large curvature path tracking accuracy. • ELOS, ILOS and SGL are adopted to compare with our ISGL. Results show the advantages in large curvature path. [ABSTRACT FROM AUTHOR]

Nanoplastics have been found in marine biological habitats. There is now an urgent need to effectively detect these nanoplastics. Due to the small size of these plastics, it is challenging to directly measure the total abundance while in a suspension. In this study, a backward 120° polarized light scattering device was used to measure the polarization parameters of Phaeodactylum tricornutum , before and after the adsorption of nanoplastics. The polarization parameters were observed to be affected by the adsorption of nanoplastics, with different concentrations and sizes. Moreover, the polarization parameter X changed with time after mixing the nanoplastics with the microalgae, which showed the details of the adsorption of nanoplastics by the microalgae, and was further confirmed by a scanning electron microscope image. The simulation results demonstrated how the polarization parameter X of the scatterers (e.g. microalgae) were affected by the adsorption of the nanoplastics. The results of this work indicate that the polarization scattering method can be used to probe the concentration and size of nanoplastics adsorbed by microalgae in water. [ABSTRACT FROM AUTHOR]