Top-view Trajectories: A Pedestrian Dataset of Vehicle-Crowd Interaction from Controlled Experiments and Crowded Campus

Abstract

Predicting the collective motion of a group of pedestrians (a crowd) under the vehicle influence is essential for the development of autonomous vehicles to deal with mixed urban scenarios where interpersonal interaction and vehiclecrowd interaction (VCI) are significant. This usually requires a model that can describe individual pedestrian motion under the influence of nearby pedestrians and the vehicle. This study proposed two pedestrian trajectory datasets, CITR dataset and DUT dataset, so that the pedestrian motion models can be further calibrated and verified, especially when vehicle influence on pedestrians plays an important role. CITR dataset consists of experimentally designed fundamental VCI scenarios (front, back, and lateral VCIs) and provides unique ID for each pedestrian, which is suitable for exploring a specific aspect of VCI. DUT dataset gives two ordinary and natural VCI scenarios in crowded university campus, which can be used for more general purpose VCI exploration. The trajectories of pedestrians, as well as vehicles, were extracted by processing video frames that come from a down-facing camera mounted on a hovering drone as the recording equipment. The final trajectories of pedestrians and vehicles were refined by Kalman filters with linear point-mass model and nonlinear bicycle model, respectively, in which xy-velocity of pedestrians and longitudinal speed and orientation of vehicles were estimated. The statistics of the velocity magnitude distribution demonstrated the validity of the proposed dataset. In total, there are approximate 340 pedestrian trajectories in CITR dataset and 1793 pedestrian trajectories in DUT dataset. The dataset is available at GitHub 1 .

Pedestrian Models for Autonomous Driving Part II: High-Level Models of Human Behavior

“…Yang et al [221] pointed out that in mixed urban scenarios, intelligent vehicles (IVs) have to cope with a certain number of surrounding pedestrians. Therefore, it is necessary to understand how vehicles and pedestrians interact with each other.…”

Towards Proactive Navigation: A Pedestrian-Vehicle Cooperation Based Behavioral Model

“…The two models are estimated and validated using the VCI-CITR data-set [11]. The scenarios in the data-set include frontal and lateral crossing interactions between a group of pedestrians (7 to 10 in each simulation) and a vehicle.…”

“…First, the ground-truth values of the CC factors (R GT ) is obtained using the earlier annotations as shown in Table I. After discarding the unidentified cases, the CC factors of all the agent pairs (R CC ) are computed using (11). These continuous values are then discretized to be compared with the ground truth values: 16) and the behavior of the agent pair is predicted to be similar (SB) if R d CC = 1, and inverse (IB) if R d CC = 0.…”

“…The cooperation model is constructed based on social rules and cognitive studies [6], as well as the VCI-CITR 1 dataset which includes top-view trajectory data of pedestrians in groups under vehicle influence in a shared space [11]. The evaluation of the pedestrians cooperation is then used to predict their trajectories.…”

Mixed Road User Trajectory Extraction From Moving Aerial Videos Based on Convolution Neural Network Detection

“…As for pedestrian trajectory extraction, Yang.et al extracts the trajectory of motor vehicles and pedestrians in the aerial video using the k-nearest-neighbor algorithm to match the scale-invariant feature and generate trajectories [23]. Bian, C.et al tracks pedestrians in low altitude UAV videos via Histogram of Oriented Gradients and the Support Vector Machine [24].…”