Modeling Language for Scenario Development of Autonomous Driving Systems
Toshiaki Aoki, Takashi Tomita, Tatsuji Kawai, Daisuke Kawakami, Nobuo Chida
TL;DR
The paper tackles ambiguity and combinatorial explosion in scenario representations for autonomous driving by introducing the Car Position Diagram (CPD), a compact graphical notation that represents numerous scenarios with boxes and tokens. It formalizes CPD into propositional logic and uses SAT-based enumeration to exhaustively generate scenarios, aided by a Python/Z3-based tool (GCPD). Experiments include lane-change modeling, performance scaling, and case studies applying CPD to JAMA and ISO 34502, demonstrating both the method's usefulness for identifying collision scenarios and its limitations with very large scenario spaces. The work shows CPD as a practical foundation for scenario analysis and mining, with future directions toward data integration, query languages, and handling more complex road networks.
Abstract
Autonomous driving systems are typically verified based on scenarios. To represent the positions and movements of cars in these scenarios, diagrams that utilize icons are typically employed. However, the interpretation of such diagrams is typically ambiguous, which can lead to misunderstandings among users, making them unsuitable for the development of high-reliability systems. To address this issue, this study introduces a notation called the car position diagram (CPD). The CPD allows for the concise representation of numerous scenarios and is particularly suitable for scenario analysis and design. In addition, we propose a method for converting CPD-based models into propositional logic formulas and enumerating all scenarios using a SAT solver. A tool for scenario enumeration is implemented, and experiments are conducted on both typical car behaviors and international standards. The results demonstrate that the CPD enables the concise description of numerous scenarios, thereby confirming the effectiveness of our scenario analysis method.