In-vehicle Navigation Technology Considering Traffic Safety: A Systematic Literature Review

With socioeconomic development, traffic safety receives increasing attention. This trend has driven traditional in-vehicle navigation technologies to shift from a single efficiency-oriented objective (minimizing travel time) toward a comprehensive optimization of traffic efficiency and safety. However, existing studies still suffer from limitations such as incomplete consideration of data dimensions, low adaptability to user needs, and difficulties in multi-objective trade-offs, which hinder their applicability in large-scale scenarios under future intelligent transportation systems. To address these gaps, this study adopts a systematic literature review approach and analyzes 51 core non-review publications. It focuses on four key research questions: data sources, traffic safety level measurement and prediction methods, safety-aware routing approaches, and technical validation strategies. The review synthesizes the current state of research and identifies core demands in the field, providing references and recommendations for the future development of navigation technologies in intelligent transportation environments. The findings indicate that, in terms of data sources, most existing studies rely on a single data source and fail to adequately capture real-time and microscopic traffic characteristics. Regarding safety level measurement and prediction, many approaches lack differentiation among various types of road units and do not sufficiently explore the evolutionary trends of traffic operational characteristics; moreover, due to challenges in data collection and the complexity of aggregation logic, achieving objective safety quantification remains difficult. For safety-aware routing methods, most studies simplify multi-objective problems into single-objective optimization, where the determination of objective weights lacks objective foundations and sufficient dynamic adaptability; although Pareto-front-based methods avoid explicit weight assignment, they face practical challenges such as reliance on preset parameters and relatively low computational efficiency. In terms of technical validation, real-world field tests provide the highest reliability but are constrained by high costs and limited scalability; simulations and data-driven scenario replay based on real data can effectively reduce testing costs, yet they often lack real-time interactive dynamics and may deviate from actual traffic conditions; subjective evaluation can supplement user perception and feedback, but trade-offs among sample size, participant characteristics, cost, and validity must be carefully balanced. Based on these insights, future research is recommended to focus on four major directions: ① establishing key navigation-related data framework under future traffic information environments. ②Integrating real-time traffic state inference into safety level prediction. ③Incorporating heterogeneous driving styles and personalized user needs to enhance routing techniques. ④Leveraging emerging technologies such as large language models to provide intelligent interaction and decision-support capabilities for next-generation navigation systems.