Autonomous Vehicles Navigation

该数据集对在包含35辆车辆和80名行人的交通场景中，对4种不同轨迹上训练后的DQN模型进行了测试，评估了导航过程中的成功与失败率。此外，数据集还包含了导航成功与失败案例的视频证据，这为分析模型的局限性和性能提供了深入见解。该数据集涵盖了4种轨迹，总计进行了100次运行，其任务是训练和测试用于自动驾驶车辆安全导航的深度强化学习模型。This dataset evaluates the navigation success and failure rates by testing the DQN model trained on four distinct trajectories in a traffic scenario containing 35 vehicles and 80 pedestrians. Additionally, the dataset includes video evidence of both successful and failed navigation cases, which offers in-depth insights for analyzing the model's limitations and performance. Covering four trajectories with a total of 100 experimental runs, this dataset is intended for training and testing deep reinforcement learning models for safe navigation of autonomous vehicles.