Cross-dimensional formation control of second-order heterogeneous multi-agent systems

Ma, Lixiang; Wang, Yu‐Long; Fei, Minrui; Pan, Quan-Ke

doi:10.1016/j.isatra.2022.02.036

Cited by 18 publications

(14 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section, we study the challenges of efficiently allocating the sub-goals in multi-agent collaboration (Wang et al, 2022). We consider the symmetry-breaking setting as in ELIGN (Ma et al, 2022). Specifically, for Coop Nav.…”

Section: Symmetry-breaking Experimentsmentioning

confidence: 99%

“…This is particularly obvious in many real-world scenarios where extrinsic rewards are sparse and delayed, exacerbating the difficulty of MARL. Deducing the reward of each agent is crucial to encourage the agents to organize their behaviors for successful collaboration (Ma et al, 2022). While reward shaping (Ng et al, 1999) requires heavy and careful manual work, previous works are striving to cope with this challenge by assigning intrinsic rewards for each agent (Ma et al, 2022;Du et al, 2019;Jaderberg et al, 2019;Stadie et al, 2015;Iqbal & Sha, 2019b).…”

Section: Goalmentioning

confidence: 99%

“…Deducing the reward of each agent is crucial to encourage the agents to organize their behaviors for successful collaboration (Ma et al, 2022). While reward shaping (Ng et al, 1999) requires heavy and careful manual work, previous works are striving to cope with this challenge by assigning intrinsic rewards for each agent (Ma et al, 2022;Du et al, 2019;Jaderberg et al, 2019;Stadie et al, 2015;Iqbal & Sha, 2019b). LIIR (Du et al, 2019) tries to break down the extrinsic reward into individual learnable intrinsic rewards for each agent.…”

Section: Goalmentioning

confidence: 99%

“…LIIR (Du et al, 2019) tries to break down the extrinsic reward into individual learnable intrinsic rewards for each agent. ELIGN (Ma et al, 2022) encourages multi-agent cooperation through learning a shared world model among different agents. However, existing methods fail to address the dynamics of whether two agents should exhibit consistent behaviors, which results in a pitfall in multi-agent reinforcement learning.…”

Section: Goalmentioning

confidence: 99%

See 3 more Smart Citations

Temporal Inconsistency-Based Intrinsic Reward for Multi-Agent Reinforcement Learning

Sun

2023

2023 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

Learning optimal behavior policy for each agent in multi-agent systems is an essential yet difficult problem. Despite fruitful progress in multi-agent reinforcement learning, the challenge of addressing the dynamics of whether two agents should exhibit consistent behaviors is still under-explored. In this paper, we propose a new approach that enables agents to learn whether their behaviors should be consistent with that of other agents by utilizing intrinsic rewards to learn the optimal policy for each agent. We begin by defining behavior consistency as the divergence in output actions between two agents when provided with the same observation. Subsequently, we introduce dynamic consistency intrinsic reward (DCIR) to stimulate agents to be aware of others' behaviors and determine whether to be consistent with them. Lastly, we devise a dynamic scale network (DSN) that provides learnable scale factors for the agent at every time step to dynamically ascertain whether to award consistent behavior and the magnitude of rewards. We evaluate DCIR in multiple environments including Multi-agent Particle, Google Research Football and StarCraft II Micromanagement, demonstrating its efficacy.

show abstract

Section: Symmetry-breaking Experimentsmentioning

confidence: 99%

Section: Goalmentioning

confidence: 99%

Section: Goalmentioning

confidence: 99%

Section: Goalmentioning

confidence: 99%

See 2 more Smart Citations

Temporal Inconsistency-Based Intrinsic Reward for Multi-Agent Reinforcement Learning

Sun

2023

2023 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

show abstract

“…In [8], the stability of m‐D systems is investigated using a wave model approach. Cross‐dimensional formation control MASs is presented in [9].…”

Section: Introductionmentioning

confidence: 99%

Stability analysis for large multi‐agent systems: A wave model approach

Tofighi

Shafiee

2022

Electronics Letters

View full text Add to dashboard Cite

A network stability test algorithm is proposed to check the stability of large‐scale multi‐agent systems (MAS). The method employs modelling and processing techniques for analyzing distributed communications in these systems. To model the data transactions within the network the Modified Fornasini‐Marchesini (MFM) model has been employed. In addition, a wave‐like model for processing the distributed data in the system has been proposed, which can ensure the stability of the entire system. This method is more general and less restrictive than other methods.

show abstract

Distributed Consensus Control for Heterogeneous Multi-agent Systems Under Communication Faults

Wang

2023

Lecture Notes in Electrical Engineering

View full text Add to dashboard Cite

Cross-dimensional formation control of second-order heterogeneous multi-agent systems

Cited by 18 publications

References 47 publications

Temporal Inconsistency-Based Intrinsic Reward for Multi-Agent Reinforcement Learning

Temporal Inconsistency-Based Intrinsic Reward for Multi-Agent Reinforcement Learning

Stability analysis for large multi‐agent systems: A wave model approach

Distributed Consensus Control for Heterogeneous Multi-agent Systems Under Communication Faults

Contact Info

Product

Resources

About