Research on autonomous driving scenario modeling and application based on environmental perception data

A method for modeling autonomous driving scenarios based on real vehicle environment perception data is proposed. Through the parsing of high-definition maps and processing of real vehicle perception data, this method enables the generation of autonomous driving simulation scenarios that closely replicate actual traffic environments. On the Matlab/Simulink platform, a simulation system was constructed, encompassing scenario models, autonomous driving algorithm models, vehicle dynamics models, among others. In addition, an application research framework was proposed, covering scenario modeling, simulation testing, and real vehicle validation. The effectiveness of this scenario modeling method was validated using real vehicle data spanning both time and space, with verification results demonstrating a high degree of consistency between the statuses of autonomous vehicles and other traffic participants in the simulation scenarios and those in the actual traffic environment, thus indicating the fidelity of the simulation scenarios. Furthermore, optimization of autonomous driving system planning and control algorithms based on scenario models and real vehicle validation was conducted to address the specific scenario of emergency obstacle avoidance failure in real vehicle applications. Results indicated that the optimized autonomous driving algorithm significantly enhances vehicle performance in similar scenarios within highly realistic simulation scenarios, thereby improving the security of algorithm optimization and validation. The proposed method exhibits substantial practical application value.