The rapid advancement of deep learning has intensified the need for comprehensive data for use by autonomous driving algorithms. High-quality datasets are crucial for the development of effective data-driven autonomous driving solutions. Next-generation autonomous driving datasets must be multimodal, incorporating data from advanced sensors that feature extensive data coverage, detailed annotations, and diverse scene representation. To address this need, we present OmniHD-Scenes, a large-scale multimodal dataset that provides comprehensive omnidirectional high-definition data. The OmniHD-Scenes dataset combines data from 128-beam LiDAR, six cameras, and six 4D imaging radar systems to achieve full environmental perception. The dataset comprises 1501 clips, each approximately 30-s long, totaling more than 450 K synchronized frames and more than 5.85 million synchronized sensor data points. We also propose a novel 4D annotation pipeline. To date, we have annotated 200 clips with more than 514 K precise 3D bounding boxes. These clips also include semantic segmentation annotations for static scene elements. Additionally, we introduce a novel automated pipeline for generation of the dense occupancy ground truth, which effectively leverages information from non-key frames. Alongside the proposed dataset, we establish comprehensive evaluation metrics, baseline models, and benchmarks for 3D detection and semantic occupancy prediction. These benchmarks utilize surround-view cameras and 4D imaging radar to explore cost-effective sensor solutions for autonomous driving applications. Extensive experiments demonstrate the effectiveness of our low-cost sensor configuration and its robustness under adverse conditions.

Lane-changing is most common driving behavior, also an important research topic of autonomous driving. Based on China-FOT database, this paper extracts 237 lane-changing cases after following process, and TTC (Time to Collision) of the beginning moment of lane-changing behavior is selected as the parameter to analyze driver's behavior characteristics. Influence of six traffic environment factors on TTC is analyzed using K-S test and one-way ANOVA (Analysis of Variance). SVM (Support Vector Machine) is employed to categorize lane-changing cases after following process according to the gravity of the danger. The result shows that, distribution of TTC when driver ends following and begins changing lane can be fitted with lognormal curve, while concerning the normal cases, the differences between individuals are unobvious. Road type, lane changing direction, type of vehicle ahead and lighting condition are significant factors. Weather condition and velocity have no significant influence on the driver's lane-change behavior after following process, however, on average, driver changes lane obviously earlier in rainy weather. A linear boundary relating to velocity between normal lane changes and pre-crash cases is determined by IO-fold cross-validation SVM.

Analysis of driver brake behavior parameters under critical cui-in scenarios is conducted based on naturalistic driving data collected in Shanghai, China. Time headway (THW) when brake initiation is chosen to evaluate driver's timing of brake initiation. Average brake pressure change rate and maximum brake jerk are to evaluate driver brake response speed, average deceleration is to evaluate the effect of driver brake maneuver, maximum deceleration is to evaluate driver's maximum brake strength. Effects of different factors on driver brake behavior parameters are analyzed by using one-way analysis of variance and linear regression analysis. Driver brake behavior parameters in this tudy are important for the development of longitudinal control systems of automated vehicles which are suitable for Chinese people.

In this paper, brake response time of 110 near-crash cases with cyclist is researched. Cyclists include bicyclist, electric bicyclist, motorcyclist and tricyclist. This paper refers to the time interval from the moment a collision threat appears to the moment the vehicle begins to decelerate to avoid the collision as brake response time (BRT). Values of BRT range from 0.47s to 2.13s with a mean of 1.016s and a standard deviation of 0.3875s. Influence of seven factors on BRT is analyzed using one-way Analysis of Variance and path analysis. Factors include occurrence time of near-crash, visibility, number of potential threat vehicles, intersection or not, road type, moving status and velocity of the vehicle. The results show that visibility, intersection or not and number of potential threat vehicles are significant factors. Better visibility in the darkness can significantly shorten BRT. BRT decreases with the increase of potential threat vehicles. However, when there are too many (more than three) potential threat vehicles ahead, drivers show significantly longer BRT. Drivers brake significantly more lately at intersection.