This study aims to explore AR-HUD(Augmented Reality-Head up Display) visual interaction cognitive load’s prediction algorithm model and obtain the best adaptation mode of AR-HUD interface visual Interaction Design. Through immersive driving simulation experiments, a driver assistance test system was established to analyze drivers’ eye movement behavior and visual resource allocation characteristics. The driver’s attention will be less focused on the driving task and correspondingly less on elements of the driving environment, negatively affecting the recovery of cognitive resources. The focus of this study is to establish a visual cognitive load index by combining the visual intensity model and the user’s subjective cognitive load evaluation of the interface. The AR-HUD visual Interaction Design coding and visual cognitive load index are used as the input and output layers to establish a visual cognitive load prediction neural network model. The neural network model was introduced into the genetic algorithm’s fitness function. The genetic algorithm was used to obtain the optimal AR-HUD Visual Interaction Design solution in the finite solution space. Then the optimal AR-HUD visual Interaction Design was obtained. The CH Scale scale was used to assess the validation of the algorithm’s soundness.