Autonomous driving techniques have been flourishing in recent years while thirsting for huge amounts of high-quality data. However, it is difficult for real-world datasets to keep up with the pace of changing requirements due to their expensive and time-consuming experimental and labeling costs. Therefore, more and more researchers are turning to synthetic datasets to easily generate rich and changeable data as an effective complement to the real world and to improve the performance of algorithms. In this paper, we summarize the evolution of synthetic dataset generation methods and review the work to date in synthetic datasets related to single and multi-task categories for the autonomous driving perception study. We also discuss the role that synthetic datasets play in the evaluation, gap test, and positive effect of autonomous driving-related algorithm testing, especially on trustworthiness and safety aspects, and provide examples of evaluation experiments. Finally, we discuss the limitations and future directions of synthetic datasets. To the best of our knowledge, this is the first survey focusing on the application of synthetic datasets in autonomous driving. This survey also raises awareness of the problems of real-world deployment of autonomous driving technology and provides researchers with a possible solution.

In this paper, we study how to learn a harmonious deep reinforcement learning (DRL) based lane-changing strategy for autonomous vehicles without Vehicle-to-Everything (V2X) communication support. The basic framework of this paper can be viewed as a multi-agent reinforcement learning in which different agents will exchange their strategies after each round of learning to reach a zero-sum game state. Unlike cooperation driving, harmonious driving only relies on individual vehicles’ limited sensing results to balance overall and individual efficiency. Specifically, we propose a well-designed reward that combines individual efficiency with overall efficiency for harmony, instead of only emphasizing individual interests like competitive strategy. Testing results show that competitive strategy often leads to selfish lane change behaviors, anarchy of crowd, and thus the degeneration of traffic efficiency. In contrast, the proposed harmonious strategy can promote traffic efficiency in both free flow and traffic jam than the competitive strategy. This interesting finding indicates that we should take care of the reward setting for reinforcement learning-based AI robots (e.g., automated vehicles) design, when the utilities of these robots are not strictly in alignment.

With the rapid development of detecting technology, the amount and scale of detected traffic data are increasing in an unbelievable speed. This paper proposed an approach for compression of traffic network flow volume data based on principal component analysis (PCA). After pre-processing by mean filter method, all the 230,400 data points are compressed together and the PCs matrix has much smaller dimensions compared to the original data. The data are recovered with the compression ratio of 6.2 and the recovery error of 13%. In addition, this compression and recovery approach is proved to be robust to the abnormal data points such as the congestion data.

In this paper, the validity of PCA eigenspectrum based Hurst exponent estimator proposed in[J. B. Gao, Y. Cao, and J.-M. Lee, ldquoPrincipal Component Analysis of 1/f ^alpha noise,rdquo Phys. Lett. A, vol. 314, no. 5-6, pp. 392-400, 2003] for single fBm signal is proved. Moreover, how to apply this estimator for fBm signals corrupted with some other signals are discussed. Theoretical analysis and experiments show that it can also be used for 1) mixed fBm signals with different Hurst exponents, 2) fBm signals corrupted with additive Gaussian white noise when the signal-to-noise ratio (SNR) is not too small, and 3) fBm signals corrupted with additive deterministic sine/cosine signals. However, the estimation accuracy depends on the SNR value for the first two situations.