Predicting the adoption of connected autonomous vehicles: A new approach based on the theory of diffusion of innovations

On the grounds that individuals heavily rely on the information that they receive from their peers when evaluating adoption of a radical innovation, this paper proposes a new approach to forecast long-term adoption of connected autonomous vehicles (CAVs). The concept of resistance is employed to explain why individuals typically tend to defer the adoption of an innovation. We assume that there exists a social network among individuals through which they communicate based on certain frequencies. In addition, individuals can be subject to media advertisement based on certain frequencies. An individual's perceptions are dynamic and change over time as the individual is exposed to advertisement and communicates with satisfied and dissatisfied adopters. We also explicitly allow willingness-to-pay (WTP) to change as a result of peer-to-peer communication. An individual decides to adopt when (i) there is a need for a new vehicles; (ii) his/her WTP is greater than CAV price; and (iii) his/her overall impression about CAVs reaches a cutoff value. Applicability of the proposed approach is shown using a survey of employees of the University of Memphis. Our results show that the automobile fleet will be near homogenous in about 2050 only if CAV prices decrease at an annual rate of 15% or 20%. We find that a 6-month pre-introduction marketing campaign may have no significant impact on adoption trend. Marketing is shown to ignite CAV diffusion but its effect is capped. CAV market share will be close to 100% only if all adopters are satisfied with their purchases; therefore, the probability that an individual becomes a satisfied adopter plays an important role in the trend of adoption. The effect of the latter probability is more pronounced as time goes by and is also more prominent when CAV price reduces at greater rates. Some caveats may be inserted when considering the study results as the findings are subject to sample bias and data limitations.