Autonomous Vehicle and its Adoption: Challenges, Opportunities, and Future Implications

The future of mobility and adoption by the public has become a matter of general concern for commuters, drivers, automotive industry and public authorities. Future transport is expected to be dominated by self-driving and autonomous vehicles. Our extensive studies on user adoption indicates that acceptance will be determined mainly by these concerns. In this paper, we investigated user adoption and acceptance of self-driving vehicles. Our study demonstrated that the future adoption of self-driving vehicles will be affected by several inherent concerns such as security, privacy, and trust by the public. Nevertheless, the results of our review revealed that the field of self-driving vehicles will continue to generate interest due to its impact on transport and mobility. Predicting user adoption of self-driving vehicles is a novel research subject which has recently begun to gather momentum; however, predictive accuracy remains a challenge using simple statistical methods. Machine learning modelling techniques can provide better understanding of data and non-linear relationships between decision variables. From our data, we were able to predict self-driving acceptance based on user preference and inherent concerns. We applied supervised learning algorithms namely, Naïve Bayes, Random Forest, and Fuzzy Logic to predict user adoption of self-driving vehicles. We used 6 independent variables namely safety, trust, security, ethics, cost, and privacy. We evaluated each algorithm by using 5-fold cross validation technique. The algorithms were compared based on 3 outcomes: accuracy, precision and recall. Fuzzy logic was found to outperform other algorithms followed by random forest while Naïve Bayes performed lower.