Ground maneuver for front-wheel drive aircraft via deep reinforcement learning

The maneuvering time on the ground accounts for 10%–30% of their flight time, and it always exceeds 50% for short-haul aircraft when the ground traffic is congested. Aircraft also contribute significantly to emissions, fuel burn, and noise when taxiing on the ground at airports. There is an urgent need to reduce aircraft taxiing time on the ground. However, it is too expensive for airports and aircraft carriers to build and maintain more runways, and it is space-limited to tow the aircraft fast using tractors. Autonomous drive capability is currently the best solution for aircraft, which can save the maneuver time for aircraft. An idea is proposed that the wheels are driven by APU-powered (auxiliary power unit) motors, APU is working on its efficient point; consequently, the emissions, fuel burn, and noise will be reduced significantly. For Front-wheel drive aircraft, the front wheel must provide longitudinal force to tow the plane forward and lateral force to help the aircraft make a turn. Forward traction effects the aircraft’s maximum turning ability, which is difficult to be modeled to guide the controller design. Deep reinforcement learning provides a powerful tool to help us design controllers for black-box models; however, the models of related works are always simplified, fixed, or not easily modified, but that is what we care about most. Only with complex models can the trained controller be intelligent. High-fidelity models that can easily modified are necessary for aircraft ground maneuver controller design. This paper focuses on the maneuvering problem of front-wheel drive aircraft, a high-fidelity aircraft taxiing dynamic model is established, including the 6-DOF airframe, landing gears, and nonlinear tire force model. A deep reinforcement learning based controller was designed to improve the maneuver performance of front-wheel drive aircraft. It is proved that in some conditions, the DRL based controller outperformed conventional look-ahead controllers.     

深空探测器总体技术

提出了深空探测器的总体设计要点,分析了总体设计约束,以及深空探测任务对探测器需求特点。基于技术的继承性和衔接性,提出了未来我国火星、小行星、金星以及木星等深空探测器的总体方案构想,讨论了任务目标、总体指标、构型和飞行过程等。研究对我国未来深空探测任务的方案设计有一定的参考价值。     

频域波束合成的深空宽带阵列信号处理

深空通信已进入了高速率传输的时代. 借助于射电天文中的小口径天线组阵思想,深空阵列网络因其优越的性价比, 已成为深空网络高速数据传输的最终解决途径.基于FX型相关器, 提出了一种集阵列信号合成、相关处理为一体的频域波束合成器架构和方法. 该架构和方法具有很强的鲁棒性和灵活性, 并可支持多任务探测. 同时,介绍了分析-综合滤波、阵列信号相干合成和自适应可变分数时延等信号处理关键技术.      

深空通信中的一种级联编码调制及其迭代接收

在深空通信中,由于通信距离大幅增加,通信信号的自由空间传播损耗很大,提高系统的功率利用效率是深空通信系统设计需要考虑的重要问题之一。提出了基于LDPC码和FQPSK调制的级联编码调制体制,首先提出该编码调制体制的系统模型,通过修改相应的算法给出适合FQPSK的迭代解调解码算法,通过计算机仿真分析了该体制的误码率性能。研究结果表明：该体制利用FQPSK的恒包络特性克服信道非线性问题并使功放工作在饱和区,利用LDPC码实现低信噪比通信,同时在接收端通过迭代解调解码获得增益,从而能够缓解深空通信中由于传输距离过远而带来的极低信噪比通信问题。     

基于改进SSD算法的航拍目标检测算法研究

研究基于深度学习技术的无人机航拍图像目标检测算法,首先介绍目标检测算法SSD(Single Shot MultiBox Detector),并对其特征提取网络进行改进,采用稠密特征提取网络替换原网络的主干特征提取网络,提高算法的特征提取能力,从而提升了算法的检测精度。针对网络实时性问题,在算法中引入分组卷积,极大地减少了网络参数量,提升了网络推理速度。为解决训练中出现的正负样本不均衡问题,利用焦点损失（Focal Loss）改进了原算法的损失函数,进一步提升了网络的收敛速度和精度。最后,通过仿真验证了改进算法在目标检测精度上的优越性。     

Encounter trajectories for deep space mission ASTER to the triple near Earth asteroid 2001-SN263. The laser altimeter (ALR) point of view

In cooperation with Russia, the Brazilian deep space mission ASTER plans to send a small spacecraft to investigate the triple asteroid 2001-SN263. The nearest launch opportunities for this project include June 2022 and June 2025. One main exploration campaign is being planned with focus on the largest asteroid (Alpha). Among the instruments under development, a laser altimeter (named ALR) was preliminarily designed and presented in 2010–2011. Many studies to define mission and instruments requirements were performed aiming at the characterization of important issues for the successful realization of the mission. Among them, the identification of a suitable trajectory that could be followed by the ASTER spacecraft in the encounter phase, when the main campaign will take place. This paper describes the effort undertaken with focus on the laser altimeter operation. Possible encounter trajectories were modelled and simulated to identify suitable approach parameters and conditions allowing the accomplishment of the intended investigation. The simulation also involves the instrument operation, considering approach geometry, attitude, relative motion, time/date, and the dynamics of the main asteroid. From the laser altimeter point of view, keeping in mind the desired coverage results (50% minimum surface coverage of asteroid Alpha, complying with horizontal and vertical resolution requirements), results point out crucial features for the encounter trajectory, like the need for a small inclination (10^-6 degrees; with respect to the asteroid's orbit), the most favourable spacecraft positioning (between the Sun and the asteroid) and pointing condition (back to the Sun), the minimum amount of achievable surface coverage (58%, focused on central areas), and the most proper time to conduct the main campaign (January 2025). Concerning the instrument, results offer refined values for divergence angle (500 to 650 μrad, half-cone), pulse repetition frequencies (from 1/20 to 1 Hz), and consequent data generation rates. A simulation tool that can use any 3D generated trajectories as input data was created for the analyses presented here. Although created for the ALR in this mission, this simple analysis tool can be adapted to other instruments in this or other missions.     

火星探测天线组阵数据接收技术研究和验证实验

为了满足我国首次火星探测任务的需要，确保探测器有效载荷科学数据的顺利下传，将采用天线组阵的方式进行数据接收。天线组阵数据接收技术在航天工程中的应用国内尚属首次，为此开展了信号合成方法和处理流程的研究，提出了利用模型计算和基于科斯塔斯环的搜索估计相结合的初始时延和多普勒频差快速估计方法。利用现有的密云站50 m天线和昆明站40 m天线，以嫦娥3号着陆器数传信号为实验对象，开展了天线组阵与数据接收技术检验实验。研究和实验结果表明，全频谱合成方法优于符号流合成方法，其合成损耗小于等于 0.45 dB ，可用于我国首次火星探测任务天线组阵的信号合成；初始时延和多普勒频差快速估计方法可提高广域组阵信号互相关的搜索效率；所确定的信号合成技术流程和数据处理方法可用于后续信号合成软硬件设备的研制和开发。     

小型核反应堆自主控制及其深空探测应用设想

针对空间小型模块化核反应堆的自主控制问题，提出自主控制体系结构，降低反应堆控制对人的依赖程度，满足了深空探测任务对空间堆自主控制的需求。首先介绍了核反应堆自主控制技术和空间探测自主技术的发展现状，分析了空间小型堆的自主控制需求，然后阐释了自主控制及核反应堆近自主控制的内涵。最后基于空间堆的运行特点，给出小型堆近自主控制分层体系结构的组成元素，并进一步建立了融合决策层和功能层的小型堆多智能体自主控制体系结构。     

Multi-block SSD based on small object detection for UAV railway scene surveillance

A method of multi-block Single Shot MultiBox Detector (SSD) based on small object detection is proposed to the railway scene of unmanned aerial vehicle surveillance. To address the limitation of small object detection, a multi-block SSD mechanism, which consists of three steps, is designed. First, the original input images are segmented into several overlapped patches. Second, each patch is separately fed into an SSD to detect the objects. Third, the patches are merged together through two stages. In the first stage, the truncated object of the sub-layer detection result is spliced. In the second stage, a sub-layer suppression and filtering algorithm applying the concept of non-maximum suppression is utilized to remove the overlapped boxes of sub-layers. The boxes that are not detected in the main-layer are retained. In addition, no sufficient labeled training samples of railway circumstance are available, thereby hindering the deployment of SSD. A two-stage training strategy leveraging to transfer learning is adopted to solve this issue. The deep learning model is preliminarily trained using labeled data of numerous auxiliaries, and then it is refined using only a few samples of railway scene. A railway spot in China, which is easily damaged by landslides, is investigated as a case study. Experimental results show that the proposed multi-block SSD method produces an overall accuracy of 96.6% and obtains an improvement of up to 9.2% compared with the traditional SSD.     

跨传感器异步迁移学习的室内单目无人机避障

针对强化学习策略由仿真环境向实际迁移困难的问题,以提高无人机采用无深度信息单目视觉时的行人规避能力为目标,提出一种基于异步深度神经网络结构的跨传感器迁移学习方法。首先,在仿真环境中仅使用虚拟单线激光雷达作为传感器,通过基于确定性策略梯度(DDPG)的深度强化学习方法,训练得到一个稳定的初级避障策略。其次,用单目摄像头和激光雷达同步采集现实环境中的视觉和深度数据集并逐帧绑定,使用上述初级避障策略对现实数据集进行自动标注,进而训练得到无需激光雷达数据的单目视觉避障策略,实现从虚拟激光雷达到现实单目视觉的跨传感器迁移学习。最后,引入YOLO v3-tiny网络与Resnet18网络组成异步深度神经网络结构,有效提高了存在行人场景下的避障性能。