Prediction of Pedestrian Spatial-Temporal Risk Levels for Intelligent Vehicles: A Data-driven Approach

In recent years, road safety has attracted significant attention from researchers and practitioners in the intelligent vehicle domain. As one of the most common and vulnerable groups of road users, pedestrians cause great concerns due to their unpredictable behaviour and movement, as subtle misunderstandings in vehicle-pedestrian interaction can easily lead to risky situations or collisions. Existing methods are usually limited by their poor generalization ability across scenarios and high demand on human calibrations. This work tackles the listed problems by proposing a Pedestrian Risk Level Prediction (PRLP) system. The system consists of three modules: data collection and processing module, pedestrian trajectory prediction module, and risk level identification module. Firstly, vehicle-perspective pedestrian data are collected. A collision-model-based risk indicator, Time-To-Collision (TTC), and spatial-temporal features, relative position and speed of pedestrian, are extracted. Using the long short-term memory model, the pedestrian trajectory prediction module predicts the relative positions in the subsequent five frames and yields speed and TTC predictions. To learn pedestrian risk patterns, a hybrid clustering and classification method is adopted to learn from the risk indicator and the spatial-temporal features, and train a risk level classifier using the learned patterns. Upon predicting the spatial-temporal features of pedestrians and identifying the corresponding risk level, the risk patterns between the ego vehicle and pedestrians are determined. Experimental results verified the capability of the PRLP system to predict the risk level of pedestrians, thus supporting the collision risk assessment of intelligent vehicles and providing safety warnings to both vehicles and pedestrians.