Custom OpenAI Gym Gridworld Environment
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/dimipan/DynamicPriorities
下载链接
链接失效反馈官方服务：
资源简介：
该数据集模拟了一个反映真实合成孔径雷达（SAR）任务动态的环境，其中包括信息优先级的变化。在这个环境中，自主代理必须按照特定的优先级顺序收集任务关键信息。此外，该环境支持动态调整信息优先级，并评估代理在优先级变化后的性能指标，如任务成功率以及优先级变化后的恢复时间。经过100个独立训练周期的测试，该数据集涵盖了多种配置。任务是在网格世界场景中进行优先级适应的信息收集。
This dataset simulates an environment that reflects the dynamics of real synthetic aperture radar (SAR) missions, including changes in information priority. In this environment, autonomous agents must collect mission-critical information following a specific priority order. Additionally, this environment supports dynamic adjustment of information priorities, and allows for the evaluation of agents' performance metrics after priority shifts, such as task success rate and recovery time following priority changes. Evaluated over 100 independent training epochs, this dataset covers multiple configurations. The task involves priority-adaptive information collection within a grid-world scenario.
提供机构：
Authors of the paper
