CoChat

A Survey of Autonomous Driving: Common Practices and Emerging Technologies

Open access

The DOI matches its Crossref record; 332 of the 359 references it cites were found in a registry or an index; 64 of the 183 sources its claims cite were checked for support.

Abstract

Automated driving systems (ADSs) promise a safe, comfortable and efficient driving experience. However, fatalities involving vehicles equipped with ADSs are on the rise. The full potential of ADSs cannot be realized unless the robustness of state-of-the-art is improved further. This paper discusses unsolved problems and surveys the technical aspect of automated driving. Studies regarding present challenges, high-level system architectures, emerging methodologies and core functions including localization, mapping, perception, planning, and human machine interfaces, were thoroughly reviewed. Furthermore, many state-of-the-art algorithms were implemented and compared on our own platform in a real-world driving setting. The paper concludes with an overview of available datasets and tools for ADS development.

Overview

CoChat Summary

TL;DR

This survey reviews automated driving systems (ADSs), covering architectures, localization/mapping, perception, assessment, planning, human-machine interaction, datasets, and tools, while also implementing and comparing state-of-the-art algorithms on the authors’ own vehicle in real-world driving. It concludes that robust urban automation and level 4/5 driving remain unsolved, with open problems in sensing, assessment, connectivity, weather robustness, and HMI.

Key findings

  • ADSs promise safety and efficiency: 94% of road accidents are attributed to human errors, and annual social benefits are projected to reach nearly $800 billion by 2050; owned-to-shared vehicles are expected to reach 50:50 by 2030.
  • Most state-of-the-art ADSs are ego-only and modular; modular pipelines risk error propagation (e.g., a Tesla perception error misclassifying a white trailer as sky led to the first ADS-related fatality), while end-to-end driving lacks hard-coded safety and interpretability and has not been implemented in real-world urban scenes except limited demonstrations.
  • Localization: GPS-IMU alone does not meet performance criteria; SLAM can work anywhere but is less efficient than a priori map-based localization; point cloud matching is state-of-the-art, but map-based methods require map-making and are affected by environmental changes.
  • Perception and planning: cameras are common but illumination-sensitive; lidar is accurate under 200 m but affected by weather and size; radar is long-range and cheap but low-resolution; event cameras have microsecond response but low resolution (DAVIS240: 240 × 180, 18.5 × 18.5 µm). Deep learning dominates detection (YOLO at 45 FPS full / 155 FPS smaller; Mask R-CNN at 5 fps), while global route planning is mature and local planning, risk assessment, driving-style recognition, and HMI handover remain open.

Methods

  • Design: a structured, comprehensive survey of ADS hardware-software practices, covering present challenges, high-level system architectures, core functions (localization, mapping, perception, assessment, planning, HMI), and available datasets, software stacks, and simulators.
  • Implementation/comparison: the authors implemented and compared state-of-the-art algorithms on their own ADS-equipped platform in real-world driving, using open-source software such as Autoware and OpenPlanner, and algorithms including NDT matching, YOLOv3, Mask R-CNN, DeepLabv3, SECOND, particle-filter tracking, and a deep spatiotemporal risk-detection framework.
  • Data/sample: no human-subject sample is described; the work uses the authors’ vehicle and campus/real-world data plus existing datasets and benchmarks such as KITTI, nuScenes, COCO, Cityscapes, Oxford RobotCar, SHRP2, Comma.ai, and DDD17.

Limitations

  • Stated in the paper: robust automated driving in urban environments has not been achieved; no production vehicle is capable of level 4 or level 5 automation; no operational connected ADS exists; the optimal sensing modality is still disputed; algorithms lack accuracy and efficiency; online assessment is needed; poor road conditions and intemperate weather remain open; V2V communication is in its infancy; HMI is under-researched; end-to-end driving lacks hard-coded safety and interpretability; deep-learning planners are not widely used in real-world systems; and a priori map-based localization requires map-making and is affected by environmental changes.
  • Evident from the text: the implementation/comparison is limited to the authors’ own platform and campus/real-world setting rather than a systematic benchmark across all surveyed methods, and the survey is narrative rather than a formal systematic review.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1109/access.2020.2983149

Cites

332 of 332 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

FIGURE 1 A high level classification of automated driving system architectures.
FIGURE 1 A high level classification of automated driving system architectures.CC-BY · From the PDF
FIGURE 2 Information flow diagrams of: (a) a generic modular system, and (b) an end-to-end driving system.
FIGURE 2 Information flow diagrams of: (a) a generic modular system, and (b) an end-to-end driving system.CC-BY · From the PDF
FIGURE 3 Ricoh Tetha V panoramic images collected using our data collection platform, in Nagoya University campus. Note some distortion still remains on the periphery of the image.
FIGURE 3 Ricoh Tetha V panoramic images collected using our data collection platform, in Nagoya University campus. Note some distortion still remains on the periphery of the image.CC-BY · From the PDF
FIGURE 4 DAVIS240 events, overlayed on the image (left) and corresponding RBG image from a different camera (right), collected by our data collection platform, at a road crossing near Nagoya University. The motion of the cyclist and vehicle causes brightness changes which trigger events.
FIGURE 4 DAVIS240 events, overlayed on the image (left) and corresponding RBG image from a different camera (right), collected by our data collection platform, at a road crossing near Nagoya University. The motion of the cyclist and vehicle causes brightness changes which trigger events.CC-BY · From the PDF
FIGURE 5 The ADS equipped Prius of Nagoya University. We have used this vehicle to perform core automated driving operations.
FIGURE 5 The ADS equipped Prius of Nagoya University. We have used this vehicle to perform core automated driving operations.CC-BY · From the PDF
FIGURE 6 We used NDT matching [101], [118] to localize our vehicle in the Nagoya University campus. White points belong to the offline pre-built map and the colored ones were obtained from online scans. The objective is to find the best match between colored points and white points, thus localizing the vehicle.
FIGURE 6 We used NDT matching [101], [118] to localize our vehicle in the Nagoya University campus. White points belong to the offline pre-built map and the colored ones were obtained from online scans. The objective is to find the best match between colored points and white points, thus localizing the vehicle.CC-BY · From the PDF
FIGURE 7 Creating a 3D pointcloud map with congregation of scans. We used Autoware [122] for mapping.
FIGURE 7 Creating a 3D pointcloud map with congregation of scans. We used Autoware [122] for mapping.CC-BY · From the PDF
FIGURE 8 An urban scene near Nagoya University, with camera and lidar data collected by our experimental vehicle and object detection outputs from state-of-the-art perception algorithms. (a) A front facing camera's view, with bounding box results from YOLOv3 [128] and (b) instance segmentation results from MaskRCNN [138]. (c) Semantic segmentation masks produced by DeepLabv3 [139]. (d) The 3D Lidar data with object detection results from SECOND [140]. Amongst the four, only the 3D perception algorithm outputs range to detected objects.
FIGURE 8 An urban scene near Nagoya University, with camera and lidar data collected by our experimental vehicle and object detection outputs from state-of-the-art perception algorithms. (a) A front facing camera's view, with bounding box results from YOLOv3 [128] and (b) instance segmentation results from MaskRCNN [138]. (c) Semantic segmentation masks produced by DeepLabv3 [139]. (d) The 3D Lidar data with object detection results from SECOND [140]. Amongst the four, only the 3D perception algorithm outputs range to detected objects.CC-BY · From the PDF
FIGURE 9 Outline of a traditional method for object detection from 3D pointcloud data. Various filtering and data reduction methods are used first, followed by clustering. The resulting clusters are shown by the different colored points in the 3D lidar data of pedestrians collected by our data collection platform.
FIGURE 9 Outline of a traditional method for object detection from 3D pointcloud data. Various filtering and data reduction methods are used first, followed by clustering. The resulting clusters are shown by the different colored points in the 3D lidar data of pedestrians collected by our data collection platform.CC-BY · From the PDF
FIGURE 10 A depth image produced from synthetic lidar data, generated in the CARLA [164] simulator.
FIGURE 10 A depth image produced from synthetic lidar data, generated in the CARLA [164] simulator.CC-BY · From the PDF
FIGURE 11 Bird's eye view perspective of 3D lidar data, a sample from the KITTI dataset [177].
FIGURE 11 Bird's eye view perspective of 3D lidar data, a sample from the KITTI dataset [177].CC-BY · From the PDF
FIGURE 12 A scene with several tracked pedestrians and cyclist with a basic particle filter on an urban road intersection. Past trajectories are shown in white with current heading and speed shown by the direction and magnitude of the arrow, sample collected by our data collection platform.
FIGURE 12 A scene with several tracked pedestrians and cyclist with a basic particle filter on an urban road intersection. Past trajectories are shown in white with current heading and speed shown by the direction and magnitude of the arrow, sample collected by our data collection platform.CC-BY · From the PDF
FIGURE 13 Annotating a 3D point cloud map with topological information. A large number of annotators were employed to build the map shown on the right-hand side. The point-cloud and annotated maps are available on [197].
FIGURE 13 Annotating a 3D point cloud map with topological information. A large number of annotators were employed to build the map shown on the right-hand side. The point-cloud and annotated maps are available on [197].CC-BY · From the PDF
Image 14
CC-BY · From the PDF
Image 15
CC-BY · From the PDF
Image 16
CC-BY · From the PDF
FIGURE 15 Global plan and the local paths. The annotated vector map shown in Figure 13 was utilized by the planner. We employed OpenPlanner [219], which is a graph-based planner, to illustrate a typical planning approach.
FIGURE 15 Global plan and the local paths. The annotated vector map shown in Figure 13 was utilized by the planner. We employed OpenPlanner [219], which is a graph-based planner, to illustrate a typical planning approach.CC-BY · From the PDF
Image 18
CC-BY · From the PDF
Image 19
CC-BY · From the PDF
Image 20
CC-BY · From the PDF
Image 21
CC-BY · From the PDF

Verification

Claims the paper makes

26 Valid · 65 Not valid · 82 Not checked · 129 Own results and interpretations

78 more claims · 19 with their cited sources checked

Create a free account to see all 302 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 114
  • Probably not a substantial claim. 162
  • Borderline: one of the tests for a substantial claim is unclear. 1

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Items 201–332 of 332

Research works printed but not linked: 23

  1. ross, 2017, The audi a8: The World’s first production car to achieve level 3 autonomy, IEEE Spectr, 1
  2. 2017, World population prospects the 2017 revision key findings and advance tables, Proc World Popul Prospect, 1
  3. 2016, Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles
  4. 2019, The Automotive Digital Transformation and the Economic Impacts of Existing Data Access Models
  5. 2019, 2019 Deloitte Global Automotive Consumer Study—Advanced Vehicle Technologies and Multimodal Transportation Global Focus Countries
  6. 2019, Udacity dataset
  7. 2018, OpenPilot
  8. 2018, Driveworks SDK
  9. maddern, 2014, Illumination invariant imaging: Applications in robust vision-based localisation, mapping and classification for autonomous vehicles, Proc Vis Place Recognit Changing Environ Workshop IEEE Int Conf Robot Automat (ICRA), 2, 3
  10. 2019, Apollo Auto
  11. somerville, 2018, Uber’s Use of Fewer Safety Sensors Prompts Questions After Arizona Crash Business News Reuters
  12. guizzo, 2011, How Google’s self-driving car works, IEEE Spectr, 18, 1132
  13. davies, 2018, Google's Self-Driving Car Caused Its First Crash
  14. lavrinc, 2018, This Is How Bad Self-Driving Cars Suck In The Rain
  15. mcfarland, 2019, Who's responsible when an autonomous car crashes?
  16. ackerman, 2017, Toyota’s gill pratt on self-driving cars and the reality of full autonomy, IEEE Spectr
  17. d’onfro, 2019, ‘I Hate Them’ Locals Reportedly Are Frustrated With Alphabet’s Self-Driving Cars
  18. jain, 1995, Machine Vision, 5
  19. 2016, Chris Urmson Explains Google Self-Driving Car Project
  20. 2018, Press Kit - Autopilot
  21. lee, 2019, How 10 Leading Companies Are Trying To Make Powerful Low-Cost Lidar
  22. lee, 2019, Autopilot Was Active When a Tesla Crashed Into a Truck Killing Driver
  23. 2019, E! 45 Programme For a European Traffic System With Highest Efficiency and Unprecedented Safety

Cited by

See all 1,889

Metadata

Authors
Yurtsever, E., Lambert, J., Carballo, A., & Takeda, K.
Affiliations
  • Nagoya University (JP)
  • TIER IV (Japan) (JP)
Venue
IEEE Access
Published
2020
Volume
8
Pages
58443-58469
ISSN
2169-3536 (ISSN-L)
OpenAlex
W2953303875
Cited by
1,889
References
359
Citations by year
2026: 230 · 2025: 421 · 2024: 372 · 2023: 301 · 2022: 240 · 2021: 232 · 2020: 91
FWCI
79.88
Type
journal article
Language
en
Open access
Locations
  • IEEE Access (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
  • Social Sciences › Human Factors and Ergonomics › Human-Automation Interaction and Safety
  • Physical Sciences › Engineering › Automotive Engineering › Vehicle Dynamics and Control Systems
Keywords
autonomous driving, autonomous vehicles, localization, mapping, perception, path planning, human-machine interaction, system architecture, driving safety, benchmark dataset, algorithm comparison

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.