To improve autonomous cars, the dynamic systems method is re-enacted. Due to sensor unreality in vehicles, artificial creation of the surrounding environment and objects is required. This study proposes a novel method for generating accurate scenario sensor data using limited LiDAR and video data from autonomous vehicles. A new SurfelGAN network recreates realistic camera images to identify cars and moving objects. Real-world data from the Waymo Open Dataset was used to evaluate scenarios. A new dataset also allows simultaneous analysis of two autonomous cars. The SurfelGAN model was tested using this dataset. The GAN-based method generates precise sensor data that supports obstacle detection and autonomous navigation. © 2023 Elsevier B.V., All rights reserved.