Route2Vec: Enabling Eficient Use of Driving Context through Contextualized Route Representations

Abstract

Understanding how vehicle occupants experience their journey is key to designing adaptive in-car systems. The environments they This work is licensed under a Creative Commons Attribution International 4.0 License. encounter, ranging from road types and trafc patterns to weather conditions, shape their mental and emotional states during a ride. Yet, leveraging this contextual information remains a challenge due to its heterogeneous nature, comprising diverse data types, such as categorical, numerical, and boolean values of various scales. We introduce Route2Vec, an attention-based framework that encodes variable-length sequences of route context into compact, semantically meaningful embeddings using a self-supervised learning pipeline. These fxed-size representations allow for efcient comparisons between diferent driving situations using common similarity metrics such as Euclidean distance. Through linear probing and Philipp Hallgarten, Thomas Kosch, Tobias Grosse-Puppendahl, and Enkelejda Kasneci qualitative analysis of the embedding space, we show that Route2Vec reliably captures salient, route-specifc characteristics. Route2Vec simplifes context-aware in-vehicle interaction by enabling designers to rapidly prototype intelligent in-vehicle interfaces. We make our trained models and code1 publicly available to foster research in this area.