Article
Chemistry, Analytical
Xiaoyu Xing, Shuyi Wang, Wenjing Liu
Summary: We constructed a spacecraft performance-fault relationship graph to help space robots identify and repair spacecraft faults efficiently. To improve the graph, we enhanced the Deep Deterministic Policy Gradient (DDPG) algorithm and proposed a relationship prediction method that combines representation learning reasoning with deep reinforcement learning reasoning. Using reinforcement learning, we trained an agent in the spacecraft performance-fault relationship graph to achieve optimal interaction with the environment. The agent's decision-making ability and value judgment accuracy were enhanced through a deep neural network-based complex value function and strategy function. Experimental evaluation on the control system's performance-fault relationship graph demonstrated our model's high prediction speed and accuracy in inferring the optimal relationship path between entities.
Article
Biochemical Research Methods
Fengqing Lu, Mufei Li, Xiaoping Min, Chunyan Li, Xiangxiang Zeng
Summary: This study introduces a computational framework called DLGN for generating bioactive molecules towards two specific targets. DLGN utilizes adversarial training and reinforcement learning to explore chemical spaces and encourage the generation of molecules that belong to the intersection of two bioactive compound distributions. The proposed model shows promise in generating novel compounds with high similarity to multiple bioactive datasets.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Computer Science, Information Systems
Lior Hirsch, Gilad Katz
Summary: NEON is a method for network pruning using deep reinforcement learning. It achieves higher generality and reduces overfitting by training on a large set of architectures, and it becomes more efficient through offline training and a novel reward function.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Artificial Intelligence
Rajeev Kumar Singh, Sonia Khetarpaul, Rohan Gorantla, Sai Giridhar Allada
Summary: This paper introduces a hybrid model approach (SHEG) for generating concise summaries and crisp headlines to capture readers' attention and convey important information. Experiments demonstrate that the approach effectively produces concise summaries and engaging headlines.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Engineering, Industrial
Yifan Zhou, Bangcheng Li, Tian Ran Lin
Summary: This paper introduces a hierarchical coordinated reinforcement learning (HCRL) algorithm to optimize maintenance of large-scale multicomponent systems, with agent parameters and coordination relationships designed based on system characteristics, and a hierarchical structure established according to components' structural importance measures. The effectiveness of the algorithm is confirmed through validation on different systems, outperforming other methods including deep reinforcement learning.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2022)
Article
Environmental Sciences
Jing Yu, Deying Liang, Bo Hang, Hongtao Gao
Summary: This study used deep reinforcement learning for dehazing aerial images affected by Earth's atmosphere. A clear-hazy aerial image dataset was developed and the dehazing results of various methods were compared. The most suitable method, DehazeNet, was extended to a multi-scale form and incorporated into a multi-agent deep reinforcement learning network called DRL_Dehaze. The results showed that DRL_Dehaze can automatically select the best dehazing method for multi-scale haze situations and generate good dehazing results in different ground types.
Review
Automation & Control Systems
Massimo Tipaldi, Raffaele Iervolino, Paolo Roberto Massenio
Summary: This paper presents and analyzes the use of Reinforcement Learning (RL) to solve spacecraft control problems in various application fields. It discusses the core elements of RL framework and provides guidelines for formulating spacecraft control problems. The adoption of RL in real space projects is also analyzed, highlighting the challenges and recommendations for future work in bridging the gap between academic solutions and industry needs.
ANNUAL REVIEWS IN CONTROL
(2022)
Article
Computer Science, Hardware & Architecture
Ke Wang, Chien-Ming Chen, M. Shamim Hossain, Ghulam Muhammad, Sachin Kumar, Saru Kumari
Summary: This paper presents a key technology for smart cities, a road target recognition algorithm, and designs a set of programs to assist automatic drivers, pedestrians, and visually impaired individuals in road safety or city infrastructure management.
Article
Chemistry, Multidisciplinary
Quoc-Dai Luong Tran, Anh-Cuong Le
Summary: This study addresses the issue of inadequate consideration of the relationship between utterances in conversation when generating responses. A novel method that comprehensively models the contextual information of the current utterance is proposed. Experimental results demonstrate the effectiveness of the proposed model, showing an improvement in average BLEU score by 24% and average ROUGE score by 29% compared to the baseline model.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Civil
Weibin Zhang, Chen Yan, Xiaofeng Li, Liangliang Fang, Yao-Jan Wu, Jun Li
Summary: To enhance intersections' throughput efficiency, this paper proposes an adaptive coordination control method based on multi-agent reinforcement learning. The method achieved stable performance in both simulated and real-world scenarios, effectively alleviating traffic congestion.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Engineering, Civil
Xing Liu, Cheng Qian, Wei Yu, David Griffith, Avi Gopstein, Nada Golmie
Summary: In this paper, the authors propose a deep reinforcement learning-based approach to automate network configurations in dynamic network environments such as the Internet of Vehicles (IoV). They use a collection of neural networks to convert the observations of a communication environment into key features and then train a deep Q neural network (DQN) to select optimal network configurations for vehicles in the IoV environment. They also consider both centralized and distributed training strategies and evaluate the efficacy of their approach using an IoV simulation platform.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Engineering, Marine
Shangyan Zou, Xiang Zhou, Irfan Khan, Wayne W. Weaver, Syed Rahman
Summary: Ocean wave energy is a sustainable energy source attracting research interest. A Deep Reinforcement Learning control outperforms model-based controls in wave power production and power quality.
Article
Computer Science, Artificial Intelligence
Hyo-Seok Hwang, Minhyeok Lee, Junhee Seok
Summary: In optical engineering, designing devices or systems with desired properties is important yet challenging. This paper proposes a deep reinforcement learning-based inverse design framework that utilizes a deep learning simulator to reduce training time and provides multiple design candidates to satisfy target properties.
APPLIED SOFT COMPUTING
(2022)
Article
Computer Science, Artificial Intelligence
Yan Xiong, Liang Guo, Yang Zhang, Mingxing Xu, Defu Tian, Ming Li
Summary: Thermal modeling is a critical technology in spacecraft thermal control systems. A new intelligent surrogate modeling strategy SMS-DL that uses deep learning has been proposed, along with an intelligent batch processing system for thermal analysis based on NX TMG Thermal Analysis to achieve a trade-off between high accuracy and low computational cost.
NEURAL COMPUTING & APPLICATIONS
(2022)
Letter
Automation & Control Systems
Zhengqing Han, Yintao Wang, Qi Sun
Summary: This letter proposes a distributed deep reinforcement learning (DRL) based approach to address the path following and formation control problems for underactuated unmanned surface vehicles (USVs). The deep deterministic policy gradient (DDPG) method is utilized by constructing two independent actor-critic architectures to determine the desired heading and speed command for each USV. The realistic dynamical model and the input saturation problem are taken into consideration. Radial basis function neural networks (RBF NNs) are employed to approximate the hydrodynamics and unknown external disturbances of USVs. Simulation results demonstrate that the proposed method can achieve high-level tracking control accuracy while maintaining a desired stable formation.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2023)
Article
Engineering, Aerospace
Jordan Maxwell, Andrew Harris, Hanspeter Schaub
Summary: A novel method using Coulomb forces for close-proximity formation flying is investigated in this study, which has applications in Earth sensing, space-situational awareness, and aeronomy. The method involves countering the differential drag acceleration between a leader and a follower craft using controlled Coulomb repulsion, ensuring precise separation. The study also explores the concept of controlling spacecraft pairs using a set of charged spheres, without the use of propellant.
Article
Engineering, Aerospace
Kieran Wilson, Alvaro Romero-Calvo, Miles Bengtson, Julian Hammerl, Jordan Maxwell, Hanspeter Schaub
Summary: This paper discusses the development, characterization, and capabilities of the Electrostatic Charging Laboratory for Interactions between Plasma and Spacecraft (ECLIPS) research vacuum chamber. This state-of-the-art facility allows conducting experiments related to charged astrodynamics in a space-like environment. It is equipped with various sources to provide electron, ion, and photon fluxes, as well as probes for characterization and a range of ancillary components to ensure safe operation.
Article
Astronomy & Astrophysics
John Martin, Hanspeter Schaub
Summary: This paper investigates the use of physics-informed neural networks (PINNs) to model the gravitational potential of the Earth and Moon, and finds that this method offers advantages in model compactness and computational efficiency compared to traditional analytic models.
CELESTIAL MECHANICS & DYNAMICAL ASTRONOMY
(2022)
Article
Engineering, Aerospace
Kieran Wilson, Julian Hammer, Hanspeter Schaub
JOURNAL OF SPACECRAFT AND ROCKETS
(2022)
Article
Engineering, Aerospace
Alvaro Romero-Calvo, Julian Hammerl, Hanspeter Schaub
Summary: In this study, the flux of secondary electrons generated over a spacecraft-like electrode assembly was investigated through experiments and simulations. The differential charging scenario was also studied. A computationally efficient three-dimensional particle tracing framework was introduced and validated as a diagnostic tool, providing theoretical and technical insights into the development of future electron-based touchless potential sensing technologies.
JOURNAL OF SPACECRAFT AND ROCKETS
(2022)
Article
Engineering, Aerospace
Kieran Wilson, Alvaro Romero-Calvo, Hanspeter Schaub
Summary: Electrostatic perturbations can have significant effects during terminal proximity operations in high Earth orbits. This paper presents a sampling-based method to minimize the impact of these perturbations and reduce the rotational rates of target objects, leading to improvements in control effort and safety.
JOURNAL OF SPACECRAFT AND ROCKETS
(2022)
Article
Engineering, Aerospace
Kieran Wilson, Miles Bengtson, Hanspeter Schaub
Summary: This study explores the fusion of two methods for remote monitoring of spacecraft electrostatic potential to generate a more accurate and robust estimate. The electron method provides high accuracy but is sensitive to target geometry, while the x-ray method is less accurate but less sensitive to geometry. By combining the two methods, significant improvements in accuracy and coverage are achieved. These results are important for future missions requiring remote monitoring of nearby objects' potential to ensure mission success.
JOURNAL OF SPACECRAFT AND ROCKETS
(2022)
Article
Engineering, Aerospace
Riccardo Calaon, Hanspeter Schaub
Summary: This paper investigates the complex attitude dynamics and control of a spacecraft under constraints, proposing a nonsingular attitude motion planning algorithm based on the Modified Rodrigues Parameter configuration space. By constructing paths with B-spline curves and imposing a constant angular maneuver rate, a constraint-compliant reference trajectory is computed to reorient the spacecraft while minimizing control effort.
JOURNAL OF SPACECRAFT AND ROCKETS
(2022)
Article
Physics, Fluids & Plasmas
Kaylee Champion, Hanspeter Schaub
Summary: This study investigates the potential of touchless potential sensing using photoemissions and secondary electron emissions for neighboring spacecraft in geosynchronous (GEO) applications. The research explores the models for electric and potential fields around a charged spacecraft in short Debye regions near the moon, and the relationship between effective Debye lengths and touchless potential sensing capabilities. The results suggest the possibility of extending this technology to cislunar regions.
IEEE TRANSACTIONS ON PLASMA SCIENCE
(2023)
Article
Engineering, Aerospace
Julian Hammerl, Andrea Lopez, Alvaro Romero-Calvo, Hanspeter Schaub
Summary: This study proposes a method to estimate the electric potential of co-orbiting spacecraft using x-rays. Experimental results show that the orientation of the target determines the areas irradiated by the electron beam and the detectability of different components. The study also introduces a new procedure to measure multiple potentials simultaneously using a single x-ray spectrum.
JOURNAL OF SPACECRAFT AND ROCKETS
(2023)
Article
Engineering, Aerospace
Anne Aryadne Bennett, Russell Carpenter, Hanspeter Schaub
Summary: Debris strikes on operational spacecraft are becoming more common, and this study aims to develop methods to detect and assess minor strikes. An extended Kalman filter with dynamic model compensation is used to estimate a spacecraft's orbit state based on simulated full-state measurements, and various test statistics are developed and compared to identify abrupt changes in spacecraft velocity. The study also performs a trade study and Monte Carlo analysis to investigate the performance of the developed techniques.
JOURNAL OF SPACECRAFT AND ROCKETS
(2023)
Article
Physics, Fluids & Plasmas
Alvaro Romero-Calvo, Kaylee Champion, Hanspeter Schaub
Summary: UV lasers are proposed as a replacement for low-energy electron beams in touchless spacecraft potential sensing due to their insensitivity to the electrostatic environment. The feasibility of this approach is verified in a representative scenario of application, and its relevance for spacecraft charge control and material identification is discussed. A simplified photoemission framework is presented and validated with vacuum chamber experiments, showing its potential for determining the spatial distribution and magnitude of photoelectrons emanating from a target surface.
IEEE TRANSACTIONS ON PLASMA SCIENCE
(2023)
Article
Engineering, Aerospace
Samuel W. Albert, Hanspeter Schaub
Summary: Motivated by cost reduction in Martian exploration, this study examines the problem of co-delivering a network of small landers to Mars in different entry trajectories. The researchers develop a linearized targeting method and nonlinear numerical optimization to design probe jettisons. Monte Carlo analyses and flight-mechanics analysis are conducted to assess the feasibility and quantify errors under uncertainties.
JOURNAL OF SPACECRAFT AND ROCKETS
(2023)
Article
Engineering, Aerospace
Samuel W. W. Albert, Hanspeter Schaub
Summary: Relative motion models are used to describe the position and velocity of a deputy spacecraft with respect to a chief spacecraft. While common models provide intuitive solutions for circular or near-circular chief orbits, their effectiveness decreases as the chief orbit becomes more eccentric. This study revisits and improves several key relative motion descriptions, reformulating them to provide an intuitive description of motion with respect to the flight path. The models are then used to estimate landing location offsets for formation flying on atmospheric entry trajectories, with predictions within 6% of the total chief range.
JOURNAL OF SPACECRAFT AND ROCKETS
(2023)
Article
Engineering, Aerospace
Adam P. Herrmann, Hanspeter Schaub
Summary: This study investigates the on-board planning for the single spacecraft, multiple ground station Earth-observing satellite scheduling problem using artificial neural network function approximation of state-action value estimates from Monte Carlo tree search. Extensive hyperparameter search is conducted for MCTS and neural network architectures to determine the best combination for data generation, with MCTS shown to compute near-optimal solutions compared to a genetic algorithm. The state-action value networks match or exceed MCTS performance in significantly less execution time, showing promise for on-board spacecraft execution.
JOURNAL OF AEROSPACE INFORMATION SYSTEMS
(2022)