Recently, deep convolutional neural networks (CNNs) have been widely explored in single image super-resolution (SISR) and contribute remarkable progress. However, most of the existing CNNs-based SISR methods do not adequately explore contextual information feature extraction stage pay little attention to final high-resolution (HR) reconstruction step, hence hindering desired SR performance. To address above two issues, this paper, we propose a two-stage attentive network (TSAN) for accurate...

This article explores deep reinforcement learning (DRL) for the flocking control of unmanned aerial vehicle (UAV) swarms. The policy is trained using a centralized-learning-decentralized-execution (CTDE) paradigm, where centralized critic network augmented with additional information about entire UAV swarm utilized to improve efficiency. Instead inter-UAV collision avoidance capabilities, repulsion function encoded as an inner-UAV "instinct." In addition, UAVs can obtain states other through...

Online accurate recognition of target tactical intention in beyond-visual-range (BVR) air combat is an important basis for deep situational awareness and autonomous decision-making, which can create pre-emptive opportunities the fighter to gain superiority. The existing methods solve this problem have some defects such as dependence on empirical knowledge, difficulty interpreting results, inability meet requirements actual combat. So online hierarchical method BVR based cascaded support...

This study deals with the autonomous evasive maneuver strategy of unmanned combat air vehicle (UCAV), which is threatened by a high-performance beyond-visual-range (BVR) air-to-air missile (AAM). Considering tactical demands achieving self-conflicting objectives in actual combat, including higher miss distance, less energy consumption and longer guidance support time, problem BVR defined reformulated into multi-objective optimization problem. Effective maneuvers UCAV used different evasion...

Abstract Background Protein–protein interactions (PPIs) are of great importance in cellular systems organisms, since they the basis structure and function many essential processes related to that. Most proteins perform their functions by interacting with other proteins, so predicting PPIs accurately is crucial for understanding cell physiology. Results Recently, graph convolutional networks (GCNs) have been proposed capture information generate representations nodes graph. In our paper, we...

For quite a long time, effective Beyond-Visual-Range (BVR) air combat tactics can only be discovered by human pilots in the actual process. However, due to lack of opportunities, making new innovation was generally considered difficult. To address this challenge, we first introduced solely end-to-end Reinforcement Learning (RL) approach for training competitive agents with adversarial self-play from scratch high fidelity simulation environment during training. Furthermore, Key Air Combat...

Deep reinforcement learning has achieved great success in laser-based collision avoidance works because the laser can sense accurate depth information without too much redundant data, which maintain robustness of algorithm when it is migrated from simulation environment to real world. However, high-cost devices are not only difficult deploy for a large scale robots but also demonstrate unsatisfactory towards complex obstacles, including irregular e.g., tables, chairs, and shelves, as well...

In the field of multiagent reinforcement learning (MARL), ability to effectively explore unknown environments and collect information experiences that are most beneficial for policy represents a critical research area. However, existing work often encounters difficulties in addressing uncertainties caused by state changes inconsistencies between agents' local observations global information, which presents significant challenges coordinated exploration among multiple agents. To address this...

The popular Proximal Policy Optimization (PPO) algorithm approximates the solution in a clipped policy space. Does there exist better policies outside of this space? By using novel surrogate objective that employs sigmoid function (which provides an interesting way exploration), we found answer is "YES", and are fact located very far from We show PPO insufficient "off-policyness", according to off-policy metric called DEON. Our explores much larger space than PPO, it maximizes Conservative...

Dogfight is often a continuous and multi-round process with missile attacks. If the fighter only considers security when evading incoming missile, it will easily lose superiority in subsequent air combat. Therefore, necessary to maintain as much tactical possible while ensuring successful evasion. The amalgamative requirements of achieving multiple evasive objectives dogfight are taken into account this paper. A method generating nondominated maneuver strategy set for missiles proposed....

Task assignment is a critical technology for heterogeneous unmanned aerial vehicle (UAV) applications. Target precedence has typically been ignored in previous studies, such that it possible to obtain task solution with an unreasonable target execution order. For this reason, cooperative multiple problem constraints (CMTAPTPC) model proposed paper, which considers not only kinematic, resource, and of the UAV, but also achieve more realistic scenarios. In addition, graph method improved...

One important component of developing autonomous agents lies in the accurate prediction their opponents' behaviors when interact with others an uncertain environment. Most recent study focuses on first constructing predictive types (or models) opponents, considering various properties interest, and subsequently using these models to predict accordingly. However, as possible type space can be rather large, it is time-consuming, sometimes even infeasible, actual opponents all candidate types....

Deep reinforcement learning has achieved great success in laser-based collision avoidance work because the laser can sense accurate depth information without too much redundant data, which maintain robustness of algorithm when it is migrated from simulation environment to real world. However, high-cost devices are not only difficult apply on a large scale but also have poor irregular objects, e.g., tables, chairs, shelves, etc. In this paper, we propose vision-based framework solve...

Relationship cognition is crucial to learning-based Multi-Robot Systems (MRSs). As an advanced application of MRSs for fierce confrontation, the relationships among autonomous air combat robots inherently present complex time-varying characteristics, which makes relationship even more difficult. However, previous studies have only focused on spatial cooperative relationships, thus ignoring potential impact temporal dynamics long-term behaviors. To tackle this drawback, we propose a novel...

Reinforcement Learning (RL) algorithms enhance intelligence of air combat Autonomous Maneuver Decision (AMD) policy, but they may underperform in target environments with disturbances. To the robustness AMD strategy learned by RL, this study proposes a Tube-based Robust RL (TRRL) method. First, introduces tube to describe reachable trajectories under disturbances, formulates method for calculating tubes based on sum-of-squares programming, and TRRL algorithm that enhances utilizing size as...

We present Coordinated Proximal Policy Optimization (CoPPO), an algorithm that extends the original (PPO) to multi-agent setting. The key idea lies in coordinated adaptation of step size during policy update process among multiple agents. prove monotonicity improvement when optimizing a theoretically-grounded joint objective, and derive simplified optimization objective based on set approximations. then interpret such CoPPO can achieve dynamic credit assignment agents, thereby alleviating...

This paper presents a three-dimensional (3D) homing guidance law against stationary target by using only bearing or angle measurement. The unobservable conditions of the 3D bearing-only relative kinematics are derived and weak observability under classical proportional navigation (PNG) is revealed. An analytical observability-enhancement considering energy consumption interception accuracy then designed based on optimal control theory in plane. To further improve system observability, we...

Artificial Intelligence (AI) has achieved a wide range of successes in autonomous air combat decision-making recently. Previous research demonstrated that AI-enabled approaches could even acquire beyond human-level capabilities. However, there remains lack evidence regarding two major difficulties. First, the existing methods with fixed decision intervals are mostly devoted to solving what act but merely pay attention when act, which occasionally misses optimal opportunities. Second, method...

Autonomous umanned aerial vehicle (UAV) manipulation is necessary for the defense department to execute tactical missions given by commanders in future unmanned battlefield. A large amount of research has been devoted improving autonomous decision-making ability UAV an interactive environment, where finding optimal maneuvering policy became one key issues enabling intelligence UAV. In this paper, we propose a algorithm air-delivery based on deep reinforcement learning under guidance expert...