Scaling Collaborative Space Networks with Deep Multi-Agent Reinforcement Learning

With the improvement in the intelligence level of UAVs and the development of cluster control technology, the intelligent decision-making method of UAV cluster confrontation will become the key technology of UAV combat in the future. The UAV cluster confrontation learning environment is of complex characteristics of high dimension, non-linearity, incomplete information, continuous action space, and so on. Recently, artificial intelligence technology represented by deep learning and reinforcement learning has made a great breakthrough. Deep reinforcement learning has shown a great ability to solve intelligent decision-making problems in complex environments. In this paper, enlightened by the multi-agent centralized training distributed execution framework and the idea of maximum policy entropy, we propose a deep reinforcement learning method of the multi-agent soft actor-critic (MASAC) with incomplete information. A game model for UAV cluster confrontation based on multi-agent deep reinforcement learning is established, and a continuous space multiple UAV combat environment is constructed. Simulation experiments are performed on the asymmetric confrontation of UAV clusters with red and blue teams. The experimental results show that MASAC outperforms the existing popular multi-agent deep reinforcement learning methods, making the game players converge to a higher return equilibrium point of the game. Moreover, the convergence of MASAC is investigated and analyzed extensively, and the results show that MASAC is of good convergence and stability, ensuring the practicability of MASAC in intelligent decision-making regarding UAV cluster confrontation.

With the increasing real-time demand in the internet era, especially dynamic requests for last-mile delivery, route planning is becoming more computationally expensive than ever before. Many supply chains (SCs) choose the joint distribution of multiple depots to reduce transportation costs and delivery times. However, providing real-time and high-quality solutions for such complex routing problems remains challenging. Current solution methods like mathematical programming and heuristics suffer from scalability issues and long computation times. In contrast, artificial intelligence, especially Deep Reinforcement Learning (DRL), provides a general-purpose framework for sequential decision-making that has produced good results for many challenging real-life problems. However, applying DRL to route multiple vehicles is nontrivial, as the joint distribution requires an effective method that facilitates collaboration and communication among all agents while they carry out the delivery mission. In this research, a collaborative Multi-Agent Deep Reinforcement Learning (MADRL) approach is proposed for routing multiple vehicles in the SC. The proposed MADRL model leverages the power of two frameworks, deep learning and reinforcement learning, to generate routing policies for all agents in real time. Experimental results show the ability of the proposed learning model to obtain fast and quality solutions for complex delivery problems. Furthermore, the generalization ability of MADRL is also validated by testing the well-trained model on different scale problems.

Attempts to address the production scheduling problem thus far rely on simplifying assumptions, such as static environment and inflexible size of the problem, which compromises the schedule performance in practice due to many unpredictable disruptions to the system. Thus, the study of scheduling in the presence of real-time events, termed dynamic scheduling, continues to attract attention given the agility, flexibility, and timeliness modern production systems must deliver. Additionally, the changing nature of the manufacturing system also raises new challenges to existing scheduling strategies. At the front-end, the development of advanced data creation and exchange frameworks such as the Internet of things and cyber-physical system and their applications to the industrial environment have created an abundance of industrial data, while at the backend, edge and cloud computing technologies greatly enhance the capacity to process that data. Industrial data must be mined and analyzed so that the investment in infrastructure is not wasted, and the production system managed more effectively and in real-time. Many data-driven technologies have been adopted in scheduling research, a promising candidate among them being reinforcement learning (RL) which is able to build a direct mapping from observation of environment to actions that improve its performance. In this thesis, a deep multi-agent reinforcement learning (deep MARL) architecture is proposed to solve the dynamic scheduling problem (DSP). The deep reinforcement learning (DRL) algorithm is used to train the decentralized scheduling agents, to capture the relationship between information on the factory floor and scheduling objectives, with the aim of making real-time decisions for a manufacturing system with frequent unexpected events. Two major aspects of deep MARL application to DSP are addressed in this work, namely the conversion from traditional static scheduling problem (SSP) to dynamic scheduling in a practical context, and the adaptation of existing deep MARL algorithms to solve the scheduling problem in such an environment. Some impractical constraints of traditional studies are removed to create a research context that is closer to actual practice, result in a scheduling problem of variable size and scope. Specialized state and action representations that can handle the ever-changing specification of problem are developed; the criteria of feature selection in dynamic environment are also discussed. Recent progressions in DRL and MARL research are integrated into the proposed approach after selection and adaptation. In addition, various improvements to common deep MARL architecture are proposed, including the lightweight multilayer perceptron (MLP) encoder that is efficient in handling unstructured industrial data, a training scheme under the multi-agent architecture to improve the stability of training and overall performance, and knowledge-based reward-shaping techniques to decompose the joint reward signal into individual utilities to speed up the learning and encourage cooperative behavior between agents. Simulation studies are then conducted for the ablation study and validation. In the first stage, the performance of the proposed approach, either as individual components or as an integrated model, are tested in iterative simulation runs within which a unique instance of production is created. Meanwhile, a set of DRL-based approaches from recent publications are run in parallel. Results suggest that the contribution of each improvement is significant; the integrated architecture also delivers stronger performance than peer DRL-based approaches. For the validation, a set of priority rules that have strong performance in specified context and are widely applied in actual production scheduling are used as the benchmark. Proposed approach also provides performance gain compared to the strongest rule, with a minor increase in computation cost and negligible latency in decision-making.

Deep reinforcement learning (DRL) has superior autonomous decision-making capabilities, combining deep learning and reinforcement learning (RL). Unlike DRL employs deep neural networks (DNNs), broad RL (BRL) adopts the broad learning system (BLS) that is established with flat networks to generate the strategy. This article proposes the multiagent adaptive broad-DRL (ABDRL) approach for traffic light control (TLC), which combines the broad network with the deep network structure. Specifically, the structure of ABDRL first expands in the form of flatted broad networks. Then, the feature representation module that contains DNNs is employed to extract the critical traffic information. In addition, experiences sampled randomly by the experience replay mechanism cannot reflect the current training status of the agent effectively. In order to alleviate the impacts caused by random sampling, the forgetful experience mechanism (FEM) is incorporated into ABDRL. The FEM enables the agent to discriminate the importance of experiences stored in the experience reply buffer to improve robustness and adaptability. We validate the effectiveness of ABDRL in TLC, and the results illustrate the optimality and robustness of ABDRL over the state-of-the-art multiagent DRL (MADRL) algorithms.

Managing large-scale, geographically distributed, and long-term risks arising from diverse underlying causes – ranging from poverty to underinvestment in protecting against natural hazards or failures of sociotechnical, economic, and financial systems – poses formidable challenges for any theory of effective social decision-making. Participants may have different and rapidly evolving local information and goals, perceive different opportunities and urgencies for actions, and be differently aware of how their actions affect each other through side effects and externalities. Six decades ago, political economist Charles Lindblom viewed “rational-comprehensive decision-making” as utterly impracticable for such realistically complex situations. Instead, he advocated incremental learning and improvement, or “muddling through,” as both a positive and a normative theory of bureaucratic decision-making when costs and benefits are highly uncertain. But sparse, delayed, uncertain, and incomplete feedback undermines the effectiveness of collective learning while muddling through, even if all participant incentives are aligned; it is no panacea. We consider how recent insights from machine learning – especially, deep multiagent reinforcement learning – formalize aspects of muddling through and suggest principles for improving human organizational decision-making. Deep learning principles adapted for human use can not only help participants in different levels of government or control hierarchies manage some large-scale distributed risks, but also show how rational-comprehensive decision analysis and incremental learning and improvement can be reconciled and synthesized.

Due to the complexity of three-dimensional (3D) human pose, it is difficult for ordinary sensors to capture subtle changes in pose, resulting in a decrease in the accuracy of 3D human pose detection. A novel 3D human motion pose detection method is designed by combining Nano sensors and multi-agent deep reinforcement learning technology. First, Nano sensors are placed in key parts of the human to collect human electromyogram (EMG) signals. Second, after de-noising the EMG signal by blind source separation technology, the time-domain and frequency-domain features of the surface EMG signal are extracted. Finally, in the multi-agent environment, the deep reinforcement learning network is introduced to build the multi-agent deep reinforcement learning pose detection model, and the 3D local pose of the human is output according to the features of the EMG signal. The fusion and pose calculation of the multi-sensor pose detection results are performed to obtain the 3D human pose detection results. The results show that the proposed method has high accuracy for detecting various human poses, and the accuracy, precision, recall and specificity of 3D human pose detection results are 0.97, 0.98, 0.95 and 0.98, respectively. Compared with other methods, the detection results in this paper are more accurate, and can be widely used in medicine, film, sports and other fields.

The rise of the artificial intelligence (AI) brings golden opportunity to accelerate the development of the intelligent transportation system (ITS). The platoon control of connected autonomous vehicle (CAV) as the key technology exhibits superior for improving traffic system. However, there still exist some challenges in multi-objective platoon control and multi-agent interaction. Therefore, this paper proposed a connected autonomous vehicle latoon control approach with multi-agent deep reinforcement learning (MADRL). Finally, the results in stochastic mixed traffic flow based on SUMO (simulation of urban mobility) platform demonstrate that the proposed method is feasible, effective and advanced.KeywordsIntelligent transportation systemConnected autonomous vehicleMulti-objective platoon controlMulti-agent deep reinforcement learning

Dear Editor, This letter presents a new transfer learning framework for the deep multi-agent reinforcement learning (DMARL) to reduce the convergence difficulty and training time when applying DMARL to a new scenario [1], [2]. The proposed transfer learning framework includes the design of neural network architecture, curriculum transfer learning (CTL) and strategy distillation. Experimental results demonstrate that our framework enables DMARL models to converge faster while improving the final performance.

A multi-agent deep reinforcement learning based energy management for behind-the-meter resources

In recent years, Positive Energy Districts (PEDs) have emerged at the forefront of urban innovation, rapidly transforming communities by integrating shared Energy Storage Systems (ESS) and Electric Vehicles (EVs) to redefine the future of sustainable communities. However, energy management in such communities remains extremely challenging due to the dynamic nature of EV availability, unpredictable renewable energy generation, and the necessity to maintain user comfort while optimizing energy use. Overcoming these challenges is critical for enabling PEDs to achieve carbon neutrality, reduce costs, and improve energy sharing. In addition, Vehicle-to-Grid (V2G) technology and shared ESS offer unique opportunities to optimize energy consumption and facilitate access to the open energy market, but fully exploiting their potential requires advanced strategies such as Deep Reinforcement Learning (DRL). To address these needs, this work proposes a novel Community Multi-Agent Deep Reinforcement Learning Vehicle-to-Grid (CoMAD V2G) solution based on Multi-Agent Reinforcement Learning (MARL), which enhances the utilization of community-generated energy and increases community autonomy by controlling the charging and discharging cycles of V2G-enabled EVs. Real data on household consumption, solar energy production, EV dynamics, and electricity prices are used to evaluate and verify the effectiveness of the proposed solution in a realistic environment. Under these conditions, the proposed solution achieves improved energy exchange with the external grid on an annual basis, a result not attained with comparable conventional heuristic or alternative learning-based approaches for the community under consideration. Furthermore, the solution reduces household electricity costs by up to 25%, highlighting its potential to deliver significant economic and sustainability benefits for PEDs.

Multi-agent deep reinforcement learning (MADRL) has been used for disaster rescue and other robotic applications. In most MADRL cases, agents are trained only to deal with specific tasks, so once there are unpredictable changes in target tasks, the earlier trained-models might no longer work, and agents have to get trained again from scratch in limited time, which means traditional MADRL is unsuited to unpredictable post-disaster environments. In order to promote the scalability and flexibility of the multi-agent system, meta-learning, which has achieved few shot learning successfully in supervised deep learning field, could reuse the earlier models as the prior knowledge to guide and speed up the new task learning process. In this work, we propose a framework to apply meta learning methods to multi-agent deep deterministic policy gradient (MADDPG), and achieve efficient adaption to new tasks with less time and fewer samples. Some common experiences in the previous tasks learning will be refined as the meta knowledge. Once the tasks get changed with new scenarios, agents will get retrained with good initial network parameters based on this meta knowledge. Although the new scenarios might be different from the previous scenarios, agents could quickly adjust their policy and get better performance within only few episodes. In two experiments, our method performs better learning efficiency than others, getting higher rewards in the first several episodes of new tasks learning.

Mobile edge computing (MEC) has emerged as a key enabler for low-latency and computation-intensive applications. However, conventional terrestrial MEC networks face challenges in coverage, scalability, and dynamic environments. To address these limitations, this paper explores unmanned aerial vehicle (UAV)-assisted MEC, leveraging the mobility and flexibility of UAVs to enhance edge computing performance. However, challenges such as dynamic mobile user association, multi-UAV collaborative trajectory design, and limited communication and computation resources remain significant hurdles. We then formulate a joint optimization problem for user association, trajectory design, and resource allocation for the multi-UAV collaborative MEC network, aiming to minimize system cost considering computation task demands and UAV resource constraints. The problem, modeled as a Mixed Integer NonLinear Program (MINLP), is reformulated as a decentralized partially observable Markov decision process (Dec-POMDP) to capture complex system dynamics. Afterwards, we propose a multi-agent generative-enhanced deep reinforcement learning method that integrates generative diffusion model (GDM) with multi-agent deep reinforcement learning (MADRL) to effectively handle high-dimensional action spaces and enable collaborative multi-UAV decision-making. Extensive simulations using a real-world mobility dataset demonstrate that the proposed method consistently outperforms three baselines in terms of time-average system cost, task completion delay, task satisfactory ratio, and energy consumption ratio when varying the number of UAVs.

When compared with their single-agent counterpart, multi-agent systems have an additional set of challenges for reinforcement learning algorithms, including increased complexity, non-stationary environments, credit assignment, partial observability, and achieving coordination. Deep reinforcement learning has been shown to achieve successful policies through implicit coordination, but does not handle partial-observability. This paper describes a deep reinforcement learning algorithm, based on multi-agent actor-critic, that simultaneously learns action policies for each agent, and communication protocols that compensate for partial-observability and help enforce coordination. We also research the effects of noisy communication, where messages can be late, lost, noisy, or jumbled, and how that affects the learned policies. We show how agents are able to learn both high-level policies and complex communication protocols for several different partially-observable environments. We also show how our proposal outperforms other state-of-the-art algorithms that don’t take advantage of communication, even with noisy communication channels.

The increasing computational demand for real-time mobile applications has led to the development of mobile edge computing (MEC), with support from unmanned aerial vehicles (UAVs), as a promising paradigm for constructing high-throughput line-of-sight links for ground users and pushing computational resources to network edges. Users can reduce processing latency and the load on their local computers by delegating tasks to the UAV in its role as an edge server. The coverage capacity of a single UAV is, however, very limited. Moreover, it will be easy to intercept the data that is transferred to the unmanned aerial vehicle. Thus, for UAV-assisted mobile edge computing, we proposed a transmission technique based on multi-agent deep reinforcement learning in this study. The recommended approach to maximize UAV deployment first applies the particle swarm optimization algorithm. Then, deep reinforcement learning is utilized to optimize the secure offloading to maximize the system utility and minimize the quantity of information eavesdropping, taking into consideration different user task types with diverse preferences for processing time and residual energy of computing equipment. The results of the simulation demonstrate that, in comparison to the single-agent strategy and the benchmark, the multi-agent approach can optimize offloading more successfully and produce higher system utility.

We propose a deep reinforcement learning algorithm for semi-cooperative multi-agent tasks, where agents are equipped with their separate reward functions, yet with some willingness to cooperate. It is intuitive that defining and directly maximizing a global reward function leads to cooperation because there is no concept of selfishness among agents. However, it may not be the best way of inducing such cooperation due to problems that arise from training multiple agents with a single reward (e.g., credit assignment). In addition, agents may intentionally be given separate reward functions to induce task prioritization whereas a global reward function may be difficult to define without diluting the effect of different tasks and causing their reward factors to be disregarded. Our algorithm, called Peer Evaluation-based Dual DQN (PED-DQN), proposes to give peer evaluation signals to observed agents, which quantify how they strategically value a certain transition. This exchange of peer evaluation among agents over time turns out to render agents to gradually reshape their reward functions so that their action choices from the myopic best response tend to result in a more cooperative joint action.