DADEL: Data-driven user evaluation of ADAS - System perspective

Purpose and goal:

Properly designed Advanced Driver Assistance Systems (ADAS, SAE L1-2) can improve traffic safety and efficiency, but this requires understanding real-world driver interaction with ADAS. Despite extensive data from connected vehicles, the automotive industry largely lacks this understanding. To address this gap and enable ADAS designers to make informed user-centered decisions, this pre-study aims to explore the integration of real-world fleet data into ADAS user evaluation by means of data-driven methods relying on techniques such as ML, data mining and predictive modeling.

Expected results and effects:

Through literature surveys, interviews, and workshops with researchers and designers, we will: a) identify what insights about ADAS usage are needed by ADAS designers and what data is currently collected from the real-world ADAS usage (with Scania as a case study), b) propose initial ideas on leveraging data-driven methods to extract relevant insights from such data, and c) outline a direction for a full-scale project where a proof-of-concept is developed to advance methods for user-centric ADAS evaluation.

Approach and implementation:

There are three work-packages (WP1-3). WP1 deals with project management. WP2 investigates what knowledge about real-world usage of ADAS is needed to optimize user evaluation and design of ADAS.

Using Scania as a use case, we will create an inventory of relevant data that is being collected from vehicles and investigate relevant data-driven methods and tools. This will inform WP3, where we aim to conduct a gap analysis between the needs and possibilities and establish an initial understanding of how data-driven methods could be utilized in evaluation and design processes of ADAS.