Learn-to-Race (L2R) Task 2.0 Benchmark
收藏arXiv2022-05-11 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2205.02953v2
下载链接
链接失效反馈官方服务：
资源简介：
Learn-to-Race (L2R) Task 2.0 Benchmark 是一个用于评估自主赛车技术的高保真模拟环境。该数据集旨在通过两阶段过程评估强化学习代理在多模态感知上的联合安全性、性能和泛化能力。第一阶段评估代理在遵守安全约束的同时尽可能快速驾驶的能力，第二阶段则要求代理通过安全探索适应未见过的赛道。数据集支持多种传感器配置和模拟赛车，旨在推动自主技术的发展，实现自动驾驶中提高安全性的承诺。
Learn-to-Race (L2R) Task 2.0 Benchmark is a high-fidelity simulation environment for evaluating autonomous racing technologies. This benchmark aims to assess the joint safety, performance, and generalization capabilities of reinforcement learning agents on multimodal perception through a two-stage process. The first stage evaluates an agent's ability to drive as fast as possible while adhering to safety constraints, and the second stage requires the agent to adapt to unseen tracks via safe exploration. Supporting multiple sensor configurations and simulated racing vehicles, this benchmark is designed to advance autonomous driving technologies and fulfill the promise of improving safety in autonomous driving.
提供机构：
卡内基梅隆大学
创建时间：
2022-05-06
