In response to increasing public transportation demands and environmental concerns, the RABus project explores the integration of automated shuttle buses (SAE Level 4) in Friedrichshafen and Mannheim to enhance urban and rural mobility. This research focuses on obstacle detection and avoidance, emphasizing the role of AI in improving automated driving safety. However, limited training data introduces AI uncertainty, affecting system reliability. Traditional safety assessments are insufficient for complex scenarios, leading to the adoption of scenario-based validation methods. This paper presents a scenario description method that combines the Operational Design Domain (ODD) and traffic behavior. By analyzing the ODD and hazard factors, a systematic approach to creating a test scenario catalog is proposed to evaluate the safety and reliability of obstacle detection systems in automated shuttle buses, addressing the need for a robust testing framework.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

A Conceptual Framework to Operational Design Domain (ODD)-Based Scenario Generation for Technical Evaluation Obstacle Detection in Automated Shuttle Bus

  • Kun Gao,
  • Ulrike Weinrich,
  • Thomas Riemer,
  • Hans-Christian Reuss

摘要

In response to increasing public transportation demands and environmental concerns, the RABus project explores the integration of automated shuttle buses (SAE Level 4) in Friedrichshafen and Mannheim to enhance urban and rural mobility. This research focuses on obstacle detection and avoidance, emphasizing the role of AI in improving automated driving safety. However, limited training data introduces AI uncertainty, affecting system reliability. Traditional safety assessments are insufficient for complex scenarios, leading to the adoption of scenario-based validation methods. This paper presents a scenario description method that combines the Operational Design Domain (ODD) and traffic behavior. By analyzing the ODD and hazard factors, a systematic approach to creating a test scenario catalog is proposed to evaluate the safety and reliability of obstacle detection systems in automated shuttle buses, addressing the need for a robust testing framework.