Using vehicle sensor data to enhance in-vehicle multimodal user experiences - How real-time fused media sensor-based data can be used to adapt and enhance multimodal in-vehicle driver and passenger experiences in terms of safety and comfort

Abstract

The advent of more advanced vehicular sensor platforms, alongside the increase in computing power, has lead to an upswing for companies working in the area called sensor fusion. Proprietary algorithms utilise vast amounts of collected data to make complex predictions otherwise too hard, or impossible, to detect. While these wonderful possibilities exist, they are according to industry leaders currently under-utilised. Among the difficulties of working with novel modality interactions, limitations such as scope, time frame and the at-the-time societal COVID-19 restrictions, are among the challenges faced by this thesis. Two things areas of focus in this thesis, which are two suggested How-to:s - How to suitably utilise fused sensor data going forward in automotive interaction design, and: How to measure what is effective about such designs. These suggested How-to:s were the hypotheses, H1 and H2, to the to the research questions. To validate the hypotheses, a design project was held following their suggestions. Parallel to the designing, evaluations (both remote and in-person) measuring task-load and usability indexes were run for potential users, i.e. drivers. Results and discussions revealed that while the first hypothesis, H1, was only validated indicatively, the thesis was successful in validating H2 - the thesis providing a fine-tuned way of measuring positive and negative impacts of such designs from early to late-stage prototypes. Furthermore, the knowledge acquired in the preparation, the learnings from the process and the formative evaluations are potentially valuable for any future work. The thesis also aimed to, and succeeded with, contributing generated ideas and artefacts to the company this thesis was done in collaboration with.