The invention discloses a method for a driverless car to predict street crossing intentions of pedestrians and relates to the technical field of driverless cars. The method comprises the following steps that: step 1, a pedestrian detection model is based on a Faster RCNN, an SE Net structure is adopted to improve the neural network convolution module on the basis, and an anchor frame width-to-height ratio with road pedestrian detection pertinence is designed through K-means clustering; 2, a particle filtering algorithm for tracking the pedestrians on a road is designed, and the pedestrians on the road are tracked by using a multi-feature fusion strategy; and step 3, a pedestrian crossing intention prediction model is designed based on the neural network. The method can enhance the prediction capability of the driverless car for the actions of the pedestrians on the road, and improves the perception and decision-making performance of the driverless car.
一种无人驾驶汽车对行人的过街意图预测方法,涉及无人驾驶汽车技术领域。步骤1:行人检测模型以Faster RCNN为基础,并在此基础上采用SE Net结构改进神经网络卷积模块,并通过K‑means聚类设计具有道路行人检测针对性的锚框宽高比;步骤2:设计对道路行人跟踪的粒子滤波算法,利用多特征融合策略实现对道路行人进行跟踪;步骤3:设计基于神经网络的行人过街意图预测模型。可增强无人驾驶汽车对道路行人动作的预测能力,提高无人驾驶汽车感知与决策性能。
Method for driverless car to predict street crossing intentions of pedestrians
一种无人驾驶汽车对行人的过街意图预测方法
2021-09-07
Patent
Electronic Resource
Chinese
Motion model estimation for pedestrians in street-crossing scenarios
Tema Archive | 2010
|Motion model estimation for pedestrians in street-crossing scenarios
Automotive engineering | 2010
|Pedestrians' Normal Walking Speed and Speed When Crossing a Street
Transportation Research Record | 2007
|Pedestrians' Normal Walking Speed and Speed When Crossing a Street
Online Contents | 2007
|