AMARL: An Attention-Based Multiagent Reinforcement Learning Approach to the Min-Max Multiple Traveling Salesmen Problem

Gao, Hao; Zhou, Xing; Xu, Xin; Lan, Yixing; Xiao, Yongqian

doi:10.1109/tnnls.2023.3236629

Cited by 6 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The advantage of this architecture is that it allows for an arbitrary number of agents compared with [15], whilst the disadvantage is also the single-depot limitation. The authors of [36] also address the single-depot mTSP and propose an attention-based multi-agent reinforcement learning (AMARL) approach that can adapt to varying numbers of agents and cities. It should be noted that a coordinator is mandatory in the architecture to avoid the interaction of agents' simultaneous decision making.…”

Section: Drl-based Methodsmentioning

confidence: 99%

“…Usually, the policy network architecture of DRL comprises an encoder to extract the deep-level features of the input and a decoder to output the action probabilities. With regard to MRTA in this paper, the intuitive input vector would be the raw coordinates of the nodes, like in most related works [15,16,36], but we think this type of input vector contains too much redundant data. In Figure 2, for example, the two graphs are indeed equivalent from the point of view of the graph configuration, because the right map is the shifted, scaled, and rotated version of the left map, but this will not influence the task allocation result, as MRTA concerns the relative location of each robot.…”

Section: Policy Network Architecturementioning

confidence: 99%

See 1 more Smart Citation

Scalable Multi-Robot Task Allocation Using Graph Deep Reinforcement Learning with Graph Normalization

Zhang,

Jiang,

Yang

et al. 2024

Electronics

View full text Add to dashboard Cite

Task allocation plays an important role in multi-robot systems regarding team efficiency. Conventional heuristic or meta-heuristic methods face difficulties in generating satisfactory solutions in a reasonable computational time, particularly for large-scale multi-robot task allocation problems. This paper proposes a novel graph deep-reinforcement-learning-based approach, which solves the problem through learning. The framework leverages the graph sample and aggregate concept as the encoder to extract the node features in the context of the graph, followed by a cross-attention decoder to output the probability that each task is allocated to each robot. A graph normalization technique is also proposed prior to the input, enabling an easy adaption to real-world applications, and a deterministic solution can be guaranteed. The most important advantage of this architecture is the scalability and quick feed-forward character; regardless of whether cases have a varying number of robots or tasks, single depots, multiple depots, or even mixed single and multiple depots, solutions can be output with little computational effort. The high efficiency and robustness of the proposed method are confirmed by extensive experiments in this paper, and various multi-robot task allocation scenarios demonstrate its advantage.

show abstract

Section: Drl-based Methodsmentioning

confidence: 99%

Section: Policy Network Architecturementioning

confidence: 99%