Road network structures and trajectory data are essential components of intelligent transportation systems (ITS), as they represent spatial infrastructure and temporal movement patterns, respec-tively. While road networks capture structural relationships and contextual information, trajectory data re- flects dynamic mobility behavior over time. Recent research has increasingly focused on integrating these two data sources using self-supervised and contrastive learning techniques to produce unified and meaningful representations. This paper presents a comprehensive review of joint contrastive representa- tion learning methods that model both intra-domain relationships (road–road and trajectory–trajectory) and inter-domain interac- tions (road–trajectory). Findings reported across multiple real- world mobility datasets indicate that these approaches achieve im-proved performance in tasks such as traffic prediction, route optimization, and trajectory similarity analysis compared to tra- ditional non-contrastive methods. In addition, this study examines commonly used evaluation strategies, highlights scalability and ethical challenges, and suggests future research directions for building adaptive and multimodal transportation systems.