基于眼动追踪的工程车辆造型设计用户体验预测模型

Predictive model of user experience of engineering vehicle modeling design based on eye tracking

为降低主观因素干扰,更加客观地评估产品造型的用户体验性,基于眼动数据,提出一种应用遗传算法优化BP神经网络的用户体验预测模型。将Tobii 120型眼动仪作为试验仪器,以Likert五级主观量表法作为辅助研究方法,采集了40位用户对12个工程车设计方案评价过程中的眼动数据,通过遗传算法对初始值编码,优化BP神经网络,建立眼动数据与主观评价相结合的综合评价模型。经广义线性回归模型筛选,确定了以注视时间为代表的10个眼动参数作为神经网络构建参数,随机选取35组眼动数据进行预测,结果显示该神经网络能有效预测产品造型设计用户主观体验得分,预测相对误差约5%。基于遗传算法的优化BP神经网络模型对使用眼动数据预测用户体验主观评价效果显著,可为今后产品造型的用户体验评价提供参考。

In order to reduce the subjective factor interference and to evaluate the user experience of product form more ob- jectively,a neural network model applying genetic algorithm based on eye movement data was proposed. Using Eye Tracker Tobii 120 and the Likert 5 scales subjective method,eye movement data was collected in the process of evaluating 12 design proposals by 40 users. The genetic algorithm was used to encode the initial value and optimize BP neural network. Then, the comprehen- sive evaluation model based on eye movement data and subjective evaluation was established. Through the selection of the gen- eralized linear regression model, 10 eye motion parameters ,which were represented by fixation time ,were used as the parameters of the neural network. 35 groups of eye movement data were randomly chosen to predict the results. The results showed that the neural network with the eye movement data can predict subjective evaluation of user experience, and the prediction relative error is about 5%. Using eye movement data,the BP neural network model based on genetic algorithm is able to predict subjective e- valuation of user experience effectively. It can provide a new method for future product user experience evaluation.