Realistic Trajectory Generation using Dynamic Deduction for Stochastic Microscopic Traffic Flow Model

The application trend of automated driving is gaining significant concern, making it increasingly crucial to validate automated driving within the stochastic simulated traffic flow environment from both time and cost perspectives. The stochastic traffic flow model attempts to encapsulate the variability inherent in traffic conditions through a stochastic process. This approach is particularly important as it accounts for the unpredictable nature of traffic, which is often not fully captured by traditional deterministic testing scenarios. However, while stochastic traffic flow models have made strides in simulating the behavior of traffic participants, there remains a significant oversight in the simulation of vehicles’ driving trajectories, leading to unrealistic portrayals of their behaviors. The trajectories of vehicles are a critical component in the overall behavior of traffic flow, and their accurate representation is essential for the simulation to reflect real-world driving patterns. This paper introduces a method that deduces the parameters in trajectory generation for stochastic microscopic traffic flow models from kinematics and dynamics, as well as the physical restriction of traffic participants. The bicycle model serves as a physical model subject to kinematic and dynamic constraints, and the resultant parameterized constraints are meticulously analyzed. A trajectory generation approach has been proposed. The proposed approach not only addresses the current limitations of stochastic traffic flow models but also paves the way for more advanced and comprehensive traffic flow simulations. It endeavors to achieve a more realistic and precise traffic flow simulation environment for automated driving.