Search

Most of the existing MAPF methods based on communication learning have poor scalability or aggregate too much redundant information, resulting in inefficient communication. To solve these problems, this paper proposed disruptor identifiable communication (DIC), which learned concise communication excluding non-disruptors by judging whether the agent in the center of the field of view would change its decision-making due to the presence of neighbors, and successfully filtered out redundant information. At the same time, this paper further instantiated DIC and developed a new highly scalable distributed MAPF solver: disruptor identifiable communication based on reinforcement and imitation learning algorithm (DICRIA). Firstly, the disruptor discriminator and the policy output layer of DICRIA identified the disruptor. Secondly, the algorithm updated the information of the disruptor and the communication wish sender in two rounds of communication respectively. Finally, DICRIA output the final policy according to the coding results of each module. Experimental results show that DICRIA S performance is better than other similar solvers in almost all environment settings, and the algorithm increases the success rate by 5.2% on average compared to the baseline solver. Especially in dense problem instances with large-size maps, the algorithm even increases the success rate of DICRIA by 44.5% compared to the baseline solver. [ABSTRACT FROM AUTHOR]

In a cooperative Markov game, each intelligence not only has to achieve a common goal, but also needs to ensure that the joint actions can satisfy the set constraints. Therefore, this paper proposed a cooperative multi-agent TD3 algorithm（MACTD3） under security constraints. Firstly, it used the attention mechanism to coordinate the actions of individual intelligences with the decision process constraints. Then it constructed a modified cost function using Lagrange multipliers. Further, in order to ensure the convergence of the algorithm, it was ensured that each intelligent body could satisfy the pre-defined constraints with different time-scale sub-learning policies. The parameters of the Actor-Critic network perform gradient descent while Lagrangian parameters on the long time scale. Finally, experimental results in heterogeneous and homogeneous cooperative multi-intelligent environments show that the MACTD3 algorithm can always obtain the minimum penalty cost compared with other algorithms. The scalability experiments by number show that MACTD3 can still satisfy the constraints with different numbers of intelligences, proving the effectiveness and scalability of the algorithm. [ABSTRACT FROM AUTHOR]

In order to overcome the challenges of policy learning in Multi-Agents Systems（MASs）, such as the unstable environment and the interaction of agent decisions, this paper proposes a method named Observation Relation Extraction（ORE）, which uses a complete graph to model the relationship between different parts of each agent’s observation, and takes advantage of the attention mechanism to calculate the importance of the relationship between different parts of each agent’s observation. By applying the above method to multi-agent reinforcement learning algorithms based on value decomposition, this paper proposes multi-agent reinforcement learning algorithms based on observation relation extraction. Experimental results on StarCraft Multi-Agent Challenge（SMAC） show the proposed algorithms with ORE leads to better performance than the original algorithms in terms of both convergence speed and final performance [ABSTRACT FROM AUTHOR]

Focusing on the problem with traffic congestion under changing environmental conditions, this paper proposed a trajectory reward light (TR-light) model by combining reinforcement learning, neural network, multi-agent and traffic simulation technology to optimize the traffic at multi-intersections. This method had considerable merits in the following aspects. The traffic organization plan was formulated based on traffic lights; multi-agent reinforcement learning was used on traffic light control; regional traffic organization was optimized through the coordination of traffic lights; the agent implemented trajectory reconstruction after the execution of each behavior so as to change the vehicle travel path without changing the OD pair, and to calculate the final reward of the agent according to the plan and reconstructed trajectory. Finally, it conducted a traffic simulation experiment through SUMO. The comparison of traffic indicators verifies that the proposed model improves the smoothness of the road network and the traffic state at the multi-intersections. Experiments show that the model is feasible and effectively mitigates the traffic congestion. [ABSTRACT FROM AUTHOR]

The self-adaptive system used in the business-critical field needs to comply with strict quality requirements. During the operation of the multi-agent adaptive system, it needs to realize adaptive adjustment according to the dynamic environment and requirements. Regarding the issue above, this paper proposed a Markov-based multi-agent adaptation run-time verification. The method first converted environmental impact factors in the multi-agent system into probabilistic form, then formalized the system into a Markov model. Finally, it conducted run-time verification through model checking so that the system could regulate itself through verification results. This paper conducted experiments through real cases of intelligent unmanned parking lots, and dis cussed the use process of the method and the verification effect. The results show that the method is suitable for adaptive systems with multi-agents, and can be adjusted in time when the system fails, and greatly improves the stability of the original system. [ABSTRACT FROM AUTHOR]

This paper studied a new event-triggered consensus protocol for MAS based on switching topology.It designed two events in this protocol,including convergence events caused by agent state evolution and topology events caused by topology change.It gave stability analysis of public Lyapunov function related to events and consensus proof of MAS.It also designed a network connectivity algorithm based on constraint set to enhance network connectivity.The simulation results show that the designed event-triggered consensus protocol can make the state evolution of agents tend to be consistent and effectively reduce the update frequency of the controller of agents in the switching topology network environment,reduce the energy consumption of MAS,improve the connectivity effect of the network effectively,and provide theoretical support for the subsequent research of event-triggered control. [ABSTRACT FROM AUTHOR]

Aiming at the problem of multi-agent cooperation in disaster environment, this paper proposed a multi-agent cooperation algorithm based on influence and state prediction. Firstly, the algorithm used the influence degree function to process the information perceived by the agent based on the information requirements of the collaborative task. Secondly, it used the predictive algorithm to predict the subsequent state of the task and the behavior of the agent, and formulated a collaborative strategy based on the prediction result. Finally, the agent performing the collaborative task dynamically adjusted the collaboration strategy according to the action effect and the trigger condition. In order to verify the effectiveness of the algorithm, it built a simulation platform in Unity 3D to compare the convergence rate, the number of rescuers, and the overall score of different collaborative algorithms. The results show that the algorithm has the faster convergence speed, the better number of rescuers and the higher of overall score, which can effectively guide the collaboration between agents and provide theoretical support for the development of practical rescue coordination strategies. [ABSTRACT FROM AUTHOR]

This paper firstly investigated the theoretical foundation and protocol framework involved in the multi-agent rendezvous problem. Secondly, it reviewed the main research results in the multi-agent rendezvous problem in recent years, and included basically the multi-agent rendezvous problem of preserving topology connectivity, improving disturbances ability and agent with resource constraints, and discussed the mainstream control methods in detail. It also summarized the multi-agent rendezvous protocol optimization, such as reducing the system power consumption and improving convergence speed. Finally, it proposed some future research directions. [ABSTRACT FROM AUTHOR]

In order to promote cooperative control theory of multi-agent based on switched topology, based on the extensive investigation of the existing literature and the latest achievements, this paper discussed the development of this field from three aspects: the consistency problem summarized the distributed optimization problem and the distributed estimation problem The status quo;such as the design of the agreement, the performance analysis of the agreement and its advantages and disadvantages, the realization of distributed optimization and the practical application of distributed estimation. Finally, it pointed out the outstanding issues in the field and the future research directions. [ABSTRACT FROM AUTHOR]

Aimed to reduce the time of the tracking system consumes when schedules tracking platforms for the dynamic highly real-time targets, this paper proposed a collaborative tracking algorithm selecting tracking platforms adaptively based on agents. Firstly, it used agents model to represent the platforms. Then, it aimed to minimize scheduling time and tracking error, and used a contract net protocol combined with binary particle swarm optimization algorithm to select the best tracking platforms. Simulation results show that the proposed algorithm can effectively reduce the scheduling time and improve the tracking accuracy compared with the existing algorithms, so it is suitable for tracking high speed moving target in real-time. [ABSTRACT FROM AUTHOR]

In current mobile core network, HSS is facing deny of service attack, man in middle attack and other security threats. Besides, in traditional static networks, attackers have asymmetric advantages over defenders. To address the problems, this paper presented a moving target defense approach of HSS based on the study of dynamic address in moving target defense. It also modeled and analysed the attacker-filter problem and proposed a fast greedy algorithm and a convex optimization method to efficiently perform the address switch strategy. The simulation results show that the fast greedy algorithm prevents attackers while significantly reducing the time complexity; the convex optimization method reduces the times of address switch byincreasing solution space and lowers the impact on normal communication of core network. [ABSTRACT FROM AUTHOR]

In order to improve the self-learning ability of traditional negotiation, this paper integrated multi-agent intelligent technology, designed the negotiation framework based on the blackboard model, constructed the five-elements negotiation model, adopted the negotiation strategy based on Q-reinforcement learning, proposed a negotiation strategy; then it optimized the negotiation strategy by the RBF neural network, predicted the information of opponent for adjusting the concession extent. At last, it verifies the feasibility and validity of the algorithm through a sample application. When comparing to the un-optimized Q-reinforcement learning, it can enhance the learning ability of the negotiation agents, reduce the negotiation time, and improve the efficiency of resolving conflicts. [ABSTRACT FROM AUTHOR]

In order to study the mechanism and influence factors of consensus evolution in mass emergency, this paper established the consensus evolution model and multi-agent simulation model which considered the social environment and individual heterogeneity characteristics. Analysis and simulation results show that there is a threshold, when the effective transmission rate is greater than it, the size of mass emergency will be stable in a certain level. Otherwise, the mass emergencies will vanish over time. Once the mass emergency broke out, the greater initial size of participate was more conducive to the spread of consensus, and the conformity would push individuals to support the dominant sides of promotion strength or preventive strength. The study will provide a meaningful reference to the prevention and control of mass emergency. [ABSTRACT FROM AUTHOR]

Aimed at the problem of synchronizing duplication in mobile database, this paper introduced the technology of multi-agent and transaction result set into synchronizing duplication of transaction level, and provided the model of synchronizing duplication in mobile database based on the technology of multi-agent. It compared with the existing models, this model had the following characteristics:a) improving the scalability of system; b) reducing the usage of mobile devices and network resources; c) proposing a series of synchronization strategy aiming at the harsh network environment. Experimental results demonstrate the excellent performance of the model and the availability in harsh network environment. [ABSTRACT FROM AUTHOR]