A Simple Rule-Based Behavioral Planner for Urban Autonomous Driving Scenarios

Autonomous driving systems require planning algorithms that can generate safe and understandable driving decisions in complex traffic environments. Although recent studies have explored advanced learning-based, adversarial, and generative methods for scenario generation and safety validation, many practical autonomous driving functions still rely on simple and interpretable planning modules. This paper presents a small-scale study of a rule-based behavioral planning framework for urban autonomous driving scenarios. The proposed framework focuses on three common driving behaviors: lane keeping, car following, and cautious yielding at intersections. Instead of using highly complex end-to-end learning, the planner uses finite-state decision logic, safety distance checking, and simple risk indicators such as time-to-collision and gap acceptance. The aim is not to outperform state-of-the-art methods, but to provide a clear and reproducible baseline that can be used in simulation-based testing and early-stage autonomous driving research. Related work on scenario-based validation, safety-critical scenario generation, crash-derived scenario analysis, and simulation platforms is reviewed to position the proposed method. The paper concludes that simple rule-based planning remains useful as a transparent baseline, especially when combined with scenario libraries and safety evaluation metrics.