Onboard observation task planning plays an essential role in satellite autonomy, which has attracted considerable attention from researchers in recent years. Most of the existing studies solve the satellite onboard observation task planning problem (SOOTP) by searching algorithms. However, the limited computing resources and the changes of onboard condition present a new challenge for these methods. In this paper, we develop a sequential decision-making model and propose a deep learning based planning method to solve the SOOTP. Instead of generating a short-term or long-term plan in advance, the sequential decision-making model enables the satellite to decide the observation task to execute in real-time. In the deep learning based planning method, a long short-term memory based encoding network is designed to extract the features and a classification network is used to make such a decision. In the experiment, we compared our method with the gated recurrent unit network and other three searching algorithms based on five scenarios. The experimental results show that our method can solve problems with 90.3%–93.7% accuracy, 2.19%– 3.95% profit gap, and 0.004s – 0.006s response time per task, which confirms its feasibility.

Road networks play a fundamental role in our daily life. It is of importance to extract the road structure in a timely and precise manner with the rapid evolution of urban road structure. Recently, road network extraction using deep learning has become an effective and popular method. The main shortcoming of the road extraction using deep learning methods lies in the fact that there is a need for a large amount of training datasets. Additionally, the datasets need to be elaborately annotated, which is usually labor-intensive and time-consuming; thus, lots of weak annotations (such as the centerline from OpenStreetMap) have accumulated over the past a few decades. To make full use of the weak annotations, we propose a novel semi-weakly supervised method based on adversarial learning to extract road networks from remote sensing imagery. Our method uses a small set of pixel-wise annotated data and a large amount of weakly annotated data for training. The experimental results show that the proposed approach can achieve a maintained performance compared with the methods that use a large number of full pixel-wise annotations while using less fully annotated data.

Semantic segmentation of remote sensing (RS) images, which is a fundamental research topic, classifies each pixel in an image. It plays an essential role in many downstream RS areas, such as land-cover mapping, road extraction, traffic monitoring, and so on. Recently, although deep-learning-based methods have shown their dominance in automatic semantic segmentation of RS imagery, the performance of these existing methods has relied heavily on large amounts of high-quality training data, which are usually hard to obtain in practice. Moreover, human-in-the-loop semantic segmentation of RS imagery cannot be completely replaced by automatic segmentation models, since automatic models are prone to error in some complex scenarios. To address these issues, in this paper, we propose an improved, smart, and interactive segmentation model, DRE-Net, for RS images. The proposed model facilitates humans’ performance of segmentation by simply clicking a mouse. Firstly, a dynamic radius-encoding (DRE) algorithm is designed to distinguish the purpose of each click, such as a click for the selection of a segmentation outline or for fine-tuning. Secondly, we propose an incremental training strategy to cause the proposed model not only to converge quickly, but also to obtain refined segmentation results. Finally, we conducted comprehensive experiments on the Potsdam and Vaihingen datasets and achieved 9.75% and 7.03% improvements in $No{C}_{95}$ compared to the state-of-the-art results, respectively. In addition, our DRE-Net can improve the convergence and generalization of a network with a fast inference speed.

This paper presents the application of a hydraulic DC-DC converter, namely a step down Buck Converter to the actuation of a robot leg that is part of the quadruped robot HyQ. The use of a Hydraulic Buck Converter (HBC) offers significant advantages in terms of improved efficiency of hydraulic actuation systems analogously to an electric switching DC-DC converter as opposed to a rheostatic-type system. In this paper, a HBC consisting of two digital valves and two check valves is introduced to improve the efficiency performance of a singl leg of a hydraulic quadruped robot (HyQ). This type of hydraulic buck converter is able to support the locomotion in two directions. The HBC operates at a switching frequency of 100 Hz in pulse-width-modulation. The better energy performance compared to proportional control is achieved by the use of fast check valves. The performance of the system with a 3-way-4-position proportional valve is compared with the HBC drive. A test rig is set up to investigate the performance of HBC with two different controllers and a Hydraulic Proportional Drive (HPD) system, based on proportional valves which control flow, by throttling it, in a dissipative manner. The performance of position tracking and energy consumption is evaluated. The experimental results indicate that HBC systems can achieve similar position tracking with relatively less consumed energy.

In this paper, an ultra-wideband 6-bit digitally controlled attenuator MMIC for the active phased array that integrates positive voltage digital driving circuits was developed, utilizing GaAs enhanced and depletion mode p-HEMT process. The attenuator design operates in an ultra-wide bandwidth of 2-18 GHz, with a dynamic attenuation range of 0.5-31.5 dB and an attenuation accuracy of less than 0.9 dB over the entire frequency band. The integrated DCFL (direct coupling field effect transistor logic) inverter circuit on the chip enables direct control of the attenuator with a positive voltage while occupying a chip area of only 2.7 mm ² .

With the development of online business, customer service agents gradually play a crucial role as an interface between the companies and their customers. Most companies spend a lot of time and effort on hiring and training customer service agents. To this end, we propose AdaCoach: A Virtual Coach for Training Customer Service Agents, to promote the ability of newly hired service agents before they get to work. AdaCoach is designed to simulate real customers who seek help and actively initiate the dialogue with the customer service agents. Besides, AdaCoach uses an automated dialogue evaluation model to score the performance of the customer agent in the training process, which can provide necessary assistance when the newly hired customer service agent encounters problems. We apply recent NLP technologies to ensure efficient run-time performance in the deployed system. To the best of our knowledge, this is the first system that trains the customer service agent through human-computer interaction. Until now, the system has already supported more than 500,000 simulation training and cultivated over 1000 qualified customer service agents.