ICRL-benchmarks
收藏资源简介:
ICRL-benchmarks是一个专为逆约束强化学习(ICRL)设计的基准数据集,涵盖了机器人控制和自动驾驶等多个应用领域的环境。该数据集由专家代理生成,用于评估ICRL算法在不同环境中的性能。数据集包括虚拟、现实和离散环境,每种环境都设计了相关的约束条件。通过这个数据集,研究者可以测试和比较不同的ICRL算法,以解决在实际应用中遇到的约束推理问题。
ICRL-benchmarks is a benchmark dataset specifically designed for Inverse Constrained Reinforcement Learning (ICRL), encompassing environments from multiple application domains such as robotic control and autonomous driving. This dataset is generated by expert agents and is used to evaluate the performance of ICRL algorithms across various environments. It includes virtual, real-world, and discrete environments, with tailored constraint conditions configured for each scenario. With this dataset, researchers can test and compare different ICRL algorithms to address constraint reasoning issues encountered in real-world application scenarios.




