A DATA-DRIVEN APPROACH TO SUPPORTING USERS’ ADAPTATION TO SMART IN-VEHICLE SYSTEMS

Doctoral thesis, 2022

The utilization of data to understand user behavior and support user needs began to develop in areas such as internet services, smartphone apps development, and the gaming industry. This bloom of data-driven services and applications forced OEMs to consider possible solutions for better in-vehicle connectivity. However, digital transformation in the automotive sector presents numerous challenges.

One of those challenges is identifying and establishing the relevant user-related data that will cover current and future needs to help the automotive industry cope with the digital transformation pace. At the same time, this development should not be sporadic, without a clear purpose or vision of how newly-generated data can support engineers to create better systems for drivers. The important issue is to learn how to extract the knowledge from the immense data we possess, and to understand the extent to which this data can be used.

Another challenge is the lack of established approaches towards vehicle data utilization for user-related studies. This area is relatively new to the automotive industry. Despite the positive examples from other fields that demonstrate the potential for data-driven context-aware applications, automotive practices still have gaps in capturing the driving context and driver behavior. This lack of user-related data can partially be explained by the multitasking activities that the driver performs while driving the car and the higher complexity of the automotive context compared to other domains. Thus, more research is needed to explore the capacity of vehicle data to support users in different tasks.

Considering all the interrelations between the driver and in-vehicle system in the defined context of use helps to obtain more comprehensive information and better understand how the system under evaluation can be improved to meet driver needs. Tracking driver behavior with the help of vehicle data may provide developers with quick and reliable user feedback on how drivers are using the system. Compared to vehicle data, the driver’s feedback is often incomplete and perception-based since the driver cannot always correlate his behavior to complex processes of vehicle performance or clearly remember the context conditions. Thus, this research aims to demonstrate the ability of vehicle data to support product design and evaluation processes with data-driven automated user insights. This research does not disregard the driver’s qualitative input as unimportant but provides insights into how to better combine quantitative and qualitative methods for more effective results.

According to the aim, the research focuses on three main aspects:

•      Identifying the extent to which vehicle data can contribute to driver behavior understanding. 

•      Expanding the concepts for vehicle data utilization to support drivers.

•      Developing the methodology for a more effective combination of quantitative (vehicle data-based) and qualitative (based on users’ feedback) studies.

Additionally, special consideration is given to describing the drawbacks and limitations, to enhance future data-driven applications.