PAsCAL D7.3 Data analysis of user acceptance from trials and simulations

The aim of PAsCAL is to develop a holistic, user-centric Guide-to- Autonomy concept aimed at accelerating the user-friendly evolution of connected and automated vehicles (CAVs) and transport systems. In doing so, it addresses important issues relating to the role of humans in this evolution, ranging from real-time driving control to long-term training needs for jobs, in particular appropriate interactions of the autonomous vehicles with different road users including disabled people and non- drivers. As part of the study, PAsCAL has conducted thorough studies in public acceptance via surveys (e.g. online, face-to-face interviews), driving scenarios in simulation as well as training and education. To validate the findings from these studies, PAsCAL carried out five real-world pilots which cover different vehicles, user groups, levels of automation, and driving experiences in a range of complex operating environments, especially as regards any problems inside and outside the autonomous vehicle, and the connectivity of transport systems with emphasis on vulnerable road users. To understand how these real-world activities were designed, planned and executed, please refer to Deliverable 6.3 – “Pilot implementation and evaluation”. This deliverable reports the results of the data analysis of these pilots, together with an online survey to reach out to the wider community as well as a use case of an autonomous air taxi to assess and elaborate common issues, approaches and lessons learned across different transport modes, with a specific focus on investigating and providing insights regarding user acceptance. The analysis was done separately per user category, namely: age, gender, willingness to pay and mobility need, where appropriate. Users’ preferences for future services and future use of CAVs and the stakeholders’ acceptance, opinions and willingness to use CAVs were studied. The data from each pilot were examined individually, as each pilot had a different focus and research questions to investigate. However, cross-pilot analysis was also conducted to better understand common “global” factors that influence acceptance. Pilot 1 “High-capacity autonomous bus operations” suggested that the absence of a human driver may be perceived as a major issue that could hinder acceptance. Moreover, unexpected malfunctioning events (doors not opening in this case) may negatively influence acceptance levels. The latter could be an indication of the heightened expectations of users with respect to the performance of automated vehicles. Given the potential negative implications of the automated technology and the absence of human staff, users may be expecting flawless performance to trust and use the technology. Actions need to be considered to compensate these negative implications of unexpected situations. Participants who felt more confident by the presence of the system were also more likely to accept the technology. To that end, enhancing confidence and trust via advanced technological communication solutions could be a future direction. Pilot 2 “Autonomous driving training” mainly focused on the effects of driver training on acceptance. With respect to that aspect, a positive attitude towards the training was correlated with positive responses towards acceptance-related questions. However, adequacy, efficiency, and whether participants would suggest similar training to others were found to have little correlation with the responses to acceptance-related questions. Vehicle reactions and difficulty to use were two additional factors that had a considerable impact on acceptance. Pilot 3 “Autonomous bus line” investigated the acceptance of a fully autonomous and connected electric bus with autonomy level 4. An interesting finding regarding this pilot was that regarding performance of the vehicle. Acceptance levels were higher when the latter was perceived as better compared to a human driver. That is, even if performance was perceived as equal, the public might decide to keep using conventional and familiar solutions, unless additional benefits from the CAV use are perceived. Pilot 4 “Shared connected transport” further highlighted the importance of vehicle performance on acceptance. Performance needs to be perceived as “more comfortable” compared to the conventional experience to ensure high acceptance levels. Similarly, the performance of the vehicle needs to be better than a human driver, as any other outcome (including performing the same as a human driver) may not suffice to ensure acceptance. Moreover, CAV reactions perceived as “very good” were significantly related to acceptance. This is a finding that also occurred in pilot 2. Finally, pilot 5 “Experience of vulnerable travellers with connected transport environment” investigated the acceptance of connected transport environment and apps. Obstacle frequency was significantly and positively correlated to two of the acceptance related questions (willingness to use and willingness to pay). In the same direction, participants who considered independence while travelling as important were also more likely to have higher acceptance levels. These are potentially the attributes that such connected environment applications should promote and highlight as their main functionalities to enhance acceptance levels. The cross-pilot analysis compared pilots using public transport modes and passenger cars. Feelings during the experience were in general significantly related to acceptance. The most commonly significant feeling expressed was that of feeling trustful. Trust is a major concept related to acceptance and it needs to be ensured in order to enhance acceptance. Another shared finding regards anticipation; participants positively surprised by the performance of CAVs were more likely to have higher levels of acceptance (even compared to those who perceived the experience “as expected”). It has been mentioned that CAVs need to outperform conventional solutions and this finding has been observed consistently across pilots. Difficulty to use and general confidence in CAVs were also consistently significant across passenger cars. Finally, it should be mentioned that participants’ background characteristics were significant only in a few cases. This suggests that the findings presented in the current document may be applicable to the average user regardless of their background; factors such as feelings, vehicle performance, safety, and difficulty to use are more important than demographic characteristics to enhance acceptance. In addition to the aforementioned pilots, Task 7.3 carried out an online survey with wider society to better understand the people’s preferences (i.e. non-users of the PAsCAL innovations) towards autonomous public transport bus/shuttle and their needs from autonomous public transport vehicles. The key conclusions obtained from the study include: ‘Using mobile applications’, ‘large fonts on HMI display’ and ‘auditory communication’ are preferred HMI methods by respondents. For requirements towards in-vehicle assistance and information, respondents prefer ‘communication with the vehicle for special assistance’, ‘visual communication with the vehicle such as HMI and mobile apps’ and the ‘presence of steward’. A high proportion of individuals responded ‘don’t know’ for improvement in passenger safety and improvement in other road safety by autonomous bus/shuttle. Finally, this deliverable also reports the study of a non-road transport use case to address common issues, approaches and lessons learned (e.g. HMI, "driver" behaviour, ethical decision making) across different transport modes. It explored attitudes to an autonomous air taxi before and after simulated flight. The work aimed to assess the current attitudes of people to autonomous air taxis and the potential to influence these attitudes via simulated flight trials. As a result of this work, the following conclusions can be drawn. Current Attitudes: People are, on the whole, positive about the concept of an autonomous air taxi. This is in terms of the safety, trust and future use of such systems. People are neutral about requiring consent from the user for the system’s actions. There is a preference for the system to have only partial control over complete control of a user journey. Correlations: There is a positive correlation between the attitudes of people and their increasing education level. There is a negative correlation between attitude and the age of the participant and driving experience There is no correlation definite correlation between the attitude of people and with the level of autonomy experience they have. Change in Attitude: The simulated flight trials provide a broadly positive increase in opinions of people towards autonomous air taxis. Using autonomy by consent provides larger increase in people’s opinion that by exception, although both modes still produce a positive increase in opinion.This publication is produced in the PAsCAL project which has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No 81509