端到端自动驾驶大模型道路交通安全法规符合性研究
马明月 1陈文华 2苗泽霖 1王韦清 1管达志3
作者信息
- 1. 公安部道路交通安全研究中心
- 2. 上海市公安局交通警察总队
- 3. 上海机动车检测认证技术研究中心有限公司
- 折叠
摘要
本文深入分析了端到端自动驾驶大模型在遵循道路交通安全法规及相关标准过程中面临的主要挑战,重点探讨了如何通过标准化的道路交通安全法规场景训练策略,提升自动驾驶大模型的法规符合性表现.最后,本文展望了未来技术发展方向,认为通过不断优化法规符合性训练和测试方法,端到端自动驾驶技术将更好地适应道路交通安全法规和相关标准,推动自动驾驶在复杂交通环境中的广泛应用.
Abstract
This paper provides a comprehensive analysis of the major challenges faced by end-to-end(E2E)autonomous driving grand model in adhering to road traffic safety law and related standards.It highlights strategies to enhance the regulatory compliance performance of these models through standardized training approaches for traffic safety scenarios.Finally,it outlines future research directions,asserting that continuous optimization and standardization of compliance training and testing methodologies will enable E2E autonomous driving technology to better align with traffic safety regulations and standards,thereby facilitating its broader application in complex traffic environments.
关键词
端到端自动驾驶大模型/道路交通安全法规/数字化交通标准规则库/实时决策验证机制Key words
E2E autonomous driving model/Road traffic safety law/Digital repository of traffic regulations/Real-time decision verification mechanisms引用本文复制引用
出版年
2024
NETL
NSTL
万方数据