Comparative Analysis of Autonomous Vehicle Simulation Environments: GTAV vs CARLA

Engineering notes

GTAV demonstrates superior realism in unpredictable human behaviors and dynamic interactions but lacks modularity.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Autonomous vehicles (AVs) require extensive simulation testing to ensure safety in complex real-world scenarios. This study presents a comparative analysis of two distinct simulation platforms: CARLA, a purpose-built AV simulator, and Grand Theft Auto V (GTAV), a commercial game adapted for AV research through DeepGTAV. We evaluate both environments using Desertic-specific driving scenarios including sandstorms, dense traffic, and extreme weather conditions. Our methodology combines quantitative metrics (lane deviation, mean speed, stopping distance) with qualitative assessments through GUI-based visualizations. Results show CARLA excels in controlled, reproducible testing with precise environmental control, achieving consistent performance metrics across scenarios. GTAV demonstrates superior realism in unpredictable human behaviors and dynamic interactions but lacks modularity. This research provides critical insights for developers selecting simulation tools for region-specific AV deployment, particularly in Middle Eastern contexts with unique environmental challenges.

Links and sources

• Official / arXiv page

Comments