A Review of Sensor Technologies for Perception in Automated Driving

Martí, Enrique; de Miguel, Miguel Ángel; García, Fernando; Rastelli, Joshué Pérez · 2019 · OpenAlex-citations

DOI: 10.1109/mits.2019.2907630

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This review paper addresses the critical role of sensor technologies in the perception systems of Automated Driving (AD) and Advanced Driver Assistance Systems (ADAS). Motivated by the high social and economic costs of traffic accidents, primarily caused by human error, the authors aim to evaluate how sensor selection and arrangement impact vehicle safety and performance. The paper focuses on exteroceptive sensors—those that perceive the vehicle’s surroundings—rather than proprioceptive sensors or communication systems. It provides a comprehensive analysis of existing and emerging technologies, linking specific sensor capabilities to perception tasks and environmental conditions, while also reviewing historical demonstrations and current commercial alliances in the field. The authors conduct a systematic review of three primary sensor technologies: artificial vision (cameras), radar, and LiDAR. For each technology, the paper details operational principles, advantages, drawbacks, and emerging variants. Artificial vision is analyzed across visible, near-infrared (NIR), and far-infrared (FIR) spectra, including 3D technologies like stereo vision, structured light, and time-of-flight (ToF). Radar technology is examined through Frequency-Modulated Continuous Wave (FMCW) systems and emerging high-resolution imaging. LiDAR is reviewed in terms of mechanical scanning systems and emerging solid-state technologies like Optical Phased Arrays (OPA). The study also establishes a taxonomy of information domains (e.g., spatial features, kinematics, semantics) and behavioral competencies to evaluate sensor suitability. This framework allows for a structured comparison of how well each sensor type acquires specific data types under various environmental conditions, such as low light, direct sunlight, rain, fog, and dust. Key findings highlight distinct trade-offs among the technologies. Cameras offer low cost and rich semantic information but struggle with varying lighting conditions and high dynamic range scenes, though NIR and FIR variants improve performance in darkness and adverse weather. Radar provides reliable long-range detection and speed measurement independent of light conditions, yet it suffers from low angular resolution and sensitivity to target reflectivity, leading to potential false positives or negatives. LiDAR delivers high-accuracy 3D point clouds and dense spatial data but is expensive, has sparse vertical resolution in lower-cost models, and is degraded by weather conditions like rain and fog. The paper notes that no single sensor is sufficient for all conditions; for instance, LiDAR may fail to detect dark, specular objects, while cameras may miss faded lane markings. Emerging technologies, such as event-based vision and FMCW LiDAR, show promise in addressing these limitations by improving dynamic range and speed detection capabilities. The significance of this work lies in its integral view of the perception pipeline, bridging raw sensor data with algorithmic requirements. By mapping sensor adequacy to specific information types and environmental factors, the paper provides a decision-making framework for designing robust AD systems. It underscores the necessity of sensor fusion to mitigate individual weaknesses and ensure reliability across diverse driving scenarios. Furthermore, the review of commercial initiatives and OEM alliances indicates a market trend toward integrated sensing solutions, highlighting the transition from research-phase demonstrations to early commercial tests. This analysis serves as a foundational reference for engineers and researchers aiming to optimize perception systems for the safety and efficacy of future automated vehicles.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• sensory abilities