The Netherlands faces many challenges regarding mobility. Strengthened by the economic recovery the number and intensity of traffic increases due to more car trips. This is accompanied by negative feedback on energy consumption, economic growth and the environment. A possible solution to cope with this problem automated driving. Automated vehicles (AVs) have the possibility to form platoons, which reduces the required space and polluted emissions. Besides, most traffic accidents are caused by human factors, which could be eliminated by a computer-driven car. An additional benefit of (full-)automated driving is that one can perform activities while driving on the road. Examples are working or having leisure time. However, it is not yet investigated how a trip using an AV as main mode is experienced compared to a trip using a conventional car. This research tries to bridge this knowledge gap by researching the following problem: ‘There is insufficient knowledge in how people will experience their trips when driving in a full-automated vehicle in relation to driving in a conventional vehicle in the Netherlands.’. A possibility to measure this perception is to determine the value of travel time savings (VOTT) of the users of automated vehicles. This has scientific value since the VOTT is used as important parameter to monetise travel time savings in cost-benefit analysis. Besides, the VOTT is used as input in traffic models. The VOTT captures a traveller’s willingness to pay for travel time savings (WTP). If the VOTT for AV users is different than for conventional car users, the importance of newly built infrastructure could change. In case of a higher VOTT, travel time savings are economically more important, while a lower VOTT reduces the importance of travel time savings and new infrastructure, ceteris paribus. The expectation is that the VOTT of AV users is lower than the VOTT of conventional car users. This expectation is based on the assumption that one is able to perform activities while driving in an AV. Because an AV user is able to work or to have leisure time instead of driving, an increase in travel time will be experienced less negatively. Thus, the aim of the research is defined as follows: ‘To explore how people in the Netherlands experience a trip in a full-automated vehicle compared to a trip in a conventional car.’. The demographic focus of this research is the Netherlands, since every country determines its own VOTTs. It is, for simplicity reasons, assumed that the AV is privately owned, and its valuation is compared to the conventional car only. This brings us to the main research question, which is: ‘How do full-automated vehicle users experience a trip compared to conventional car users for the trip purpose home-to-work in the Netherlands?’. Different methods can be used to determine VOTTs. Given the nature of the research it is chosen to combine stated preference (SP) experiments and an exploratory factor analysis (EFA). The main advantages of an SP experiment are that it is able to cope with non-existing alternatives, the VOTT can be statistically derived from discrete choice models and it allows respondents to choose between alternatives rather than rating alternatives. An exploratory factor analysis was chosen, because it was expected that psychological factors regarding automated driving influence the decision-making. The EFA will be executed by means of a latent variable model. A hybrid choice modelling approach has been applied, where the latent variable model and the discrete choice models are estimated sequentially. Two experiments were held. The first SP experiment compares two types of AVs to the conventional car. The second SP experiment substitutes the AVs for chauffeur-driven (CH) cars. At the end the experience of a trip in an AV is compared to the experience of a trip in a chauffeur- driven car. Two AV/CH variants are defined: an AV/CH with office interior and an AV/CH with leisure interior. This has been done to explore if there is a difference in trip experience when one is working or when one is having leisure time. The SP experiments explore the classical instrumental variables travel time, travel costs and walking time. Travel company [travel alone, travel with family/friends] and AV/CH-office activity [working extra time, saving time at the office] are added as additional attributes in the SP experiment. Two principles of discrete choice modelling exist: random utility maximisation (RUM) based and random regret minimisation (RRM) based. RRM models assume that respondents choose the alternative that generates least regret, while RUM models assume that respondents pick the alternative that produces most utility. It is chosen to use the RUM principle in this thesis, since it is easier to derive the VOTT estimates and the VOTT estimate is more complete. Besides, RUM is a more commonly used methodology and it is expandable with extensions like latent variable models. In total each SP experiment (AV-case and chauffeur-case) included 12 different choice sets. Each choice task included the same travelling context, which is the morning peak (from home to work). The final survey included further 18 attitudinal statements that had to be rated. The last part of the survey included questions about socio-economic characteristics of the respondents. Each survey (AV-case and chauffeur-case) was distributed through different large online panels. (Author/publisher)
Abstract