Omnisafe
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/PKU-Alignment/omnisafe.git
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是基于Safety Gym开发的一个流行的安全强化学习基准,旨在评估在安全约束下的算法表现。在Omnisafe中,除了成本约束被设定为速度限制外,机器人的速度还会决定奖励,这一点与Safety-MuJoCo基准有所不同。其任务是进行安全强化学习。
This dataset is a widely used safe reinforcement learning benchmark developed based on Safety Gym, which aims to evaluate the performance of algorithms under safety constraints. In Omnisafe, in addition to the cost constraint being set as a speed limit, the robot's speed also determines the reward, which differs from the Safety-MuJoCo benchmark. The core task of this benchmark is safe reinforcement learning.
提供机构:
PKU-Alignment



