自动驾驶中量子认知与决策问题研究

Research on quantum cognition and decision making in autonomous driving

自动驾驶车辆对人类驾驶车辆和行人的意图估计及其相互作用研究是极其重要的,现有的研究不能很好的解释人类交通参与者的不确定因素和非理性行为,这对研究自动驾驶车辆在真实道路交通场景中运行形成了阻碍,本文基于量子理论和锚定效应,针对自动驾驶车辆右转时与非机动车和行人交互场景,构建量子决策模型.仿真分析和数据集实验证明了在与人类交通参与者进行交互时,锚定效应下的量子决策模型可以考虑存在不确定性因素和非理性行为时进行加速或减速的决策,且相比于累积前景理论模型(CPT)更加贴合实际情况.

Autonomous vehicles will share roads with human traffic participants for a long time to come,it is extremely important to study the estimation of human intention to drive vehicles and pedestrians and their interactions,existing studies cannot well explain the uncertainty and irrational behavior of human traffic participants,this has hindered the study of autonomous vehicles operating in real road traffic scenarios.In this paper,based on quantum theory and anchoring effect,a quantum decision model is constructed for the interaction scene when the autonomous vehicle is turning right with non-motor vehicles and pedestrians.The simulation analysis and data set experiments show that when interacting with human traffic participants,the quantum decision model under anchoring effect can accelerate or decelerate decisions in the presence of uncertain factors and irrational behaviors,through experimental comparison of data sets show that quantum decision model is more consistent with the actual situation than the cumulative prospect theory(CPT)model.