Page 71 - Read Online

Ma et al. Intell Robot 2023;3(4):581-95 I http://dx.doi.org/10.20517/ir.2023.33 Page 11 of 15

(a) (b) (c)

Figure 3. Algorithm test results in multi-task scenarios.

Figure 4. Accuracy of task allocation algorithm under multi-agent strategy.

We conducted tests to evaluate the accuracy of multi-agent deployment strategies under different task assign-
ments using the proposed algorithms, as illustrated in Figure 4. Tasks 1 and 2 were assigned to unmanned ve-
hicles for execution, tasks 3 and 4 were assigned to UAVs for execution, and task 5 was assigned to unmanned
ships for execution. The innovative algorithm demonstrated in this article showcases improved accuracy in
allocating different tasks to different multi-agent executions.

In the node classification task, we successfully classified the agents in the multi-agent system. For example,
we categorized the UAV, unmanned ship, and unmanned vehicle into different types, such as reconnaissance
type and rescue type. This accurate classification enables intelligent agent systems to better understand and
respond to emergency events, providing robust support for decision-making.

The algorithm structure proposed in this article has an excellent mechanism for fusion recognition of global
and local features. By learning global and local features, it can focus on some main intelligent agents that are
executing tasks. Figure 5 shows the intelligent recognition of tasks of the algorithm focused on drones and