Autonomous Vehicle Routing and Navigation, Mobility Simulation and Traffic Flow Prediction Tools, and Deep Learning Object Detection Technology in Smart Sustainable Urban Transport Systems.

• Author: Poliak, Milos

1. Introduction

   Multi-sensor environmental data fusion, environment perception systems, and deep convolutional neural networks are pivotal in connected autonomous vehicles. The purpose of our systematic review is to examine the recently published literature on smart sustainable urban transport systems and integrate the insights it configures on autonomous vehicle routing and navigation, mobility simulation and traffic flow prediction tools, and deep learning object detection technology. By analyzing the most recent (2021-2022) and significant (Web of Science, Scopus, and ProQuest) sources, our paper has attempted to prove that self-driving car control algorithms leverage augmented reality capabilities, sensor and data processing technologies, and real-time predictive analytics. The actuality and novelty of this study are articulated by addressing predictive modeling and virtual simulation algorithms, ambient sound recognition software, Internet of Things connected sensors, and movement and behavior tracking tools (Andronie et al, 2021a, b; Kliestik et al, 2022; Nica, 2018), that is an emerging topic involving much interest. Our research problem is whether urban mobility data analytics harness predictive control algorithms, multi-sensor environmental data fusion, and mobile cloud and edge computing systems.

   In this review, prior findings have been cumulated indicating that block-chain and data fusion technologies and interactive data visualization tools (Balica, 2022; Konhausner et al, 2021; Lazaroiu, 2018; Olssen, 2021) enhance urban transportation systems. The identified gaps advance lane detection and simulation modeling algorithms, traffic sensing technology, and geospatial mapping technologies. Our main objective is to indicate that urban sensing technologies (de Godoy et al, 2022; Lazaroiu et al, 2020; Peters, 2022; Poliak et al, 2021) integrate cloud and edge computing algorithms, virtual simulation modeling tools, and visual analytics. This systematic review contributes to the literature on spatio-temporal fusion and object recognition algorithms, mapping and navigation tools, algorithm-driven sensing devices, and real-time Internet of Things data (Dusmanescu et al., 2016; Lazaroiu et al, 2022; Pocol et al, 2022; Pop et al, 2022) by clarifying that driving automation systems shape smart mobility.

2. Theoretical Overview of the Main Concepts

   Optimal trajectory planning, traffic management systems, connected technologies, action modeling, and driving control configure the smart road transport infrastructure. Sensor data processing and predictive control algorithms, geospatial analytics, and cognitive computing systems shape networked urban environments. The manuscript is organized as following: theoretical overview (section 2), methodology (section 3), virtual and augmented reality tools, sensing and navigation systems, and autonomous driving algorithms and behaviors (section 4), lane detection and simulation modeling algorithms, traffic sensing technology, and geospatial mapping technologies (section 5), deep learning object detection technology, autonomous vehicle decision-making algorithms, and sensor data fusion (section 6), discussion (section 7), synopsis of the main research outcomes (section 8), conclusions (section 9), limitations, implications, and further directions of research (section 10).

3. Methodology

   Throughout April 2022, a quantitative literature review of the Web of Science, Scopus, and ProQuest databases was performed, with search terms including "smart sustainable urban transport systems" + "autonomous vehicle routing and navigation," "mobility simulation and traffic flow prediction tools," and "deep learning object detection technology." The search terms were determined as being the most employed words or phrases across the analyzed literature. As research published between 2021 and 2022 was inspected, only 89 articles satisfied the eligibility criteria. By taking out controversial or ambiguous findings (insufficient/irrelevant data), outcomes unsubstantiated by replication, too general material, or studies with nearly identical titles, we selected 15 mainly empirical sources (Tables 1 and 2). Extracting and inspecting publicly accessible files (scholarly sources) as evidence, before the research began no institutional ethics approval was required. Data visualization tools: Dimensions (bibliometric mapping) and VOSviewer (layout algorithms). Reporting quality assessment tool: PRISMA. Methodological quality assessment tools include: AMSTAR, Dedoose, Distiller SR, and SRDR (Figures 1-6).

4. Virtual and Augmented Reality Tools, Sensing and Navigation Systems, and Autonomous Driving Algorithms and Behaviors

   Predictive modeling and virtual simulation algorithms, ambient sound recognition software, Internet of Things connected sensors, and movement and behavior tracking tools (Nair and Bhat, 2021; Shannon et al., 2021) cut down crashes and casualties. Cloud computing and image recognition technologies, sensor fusion algorithms, data mining and mobility simulation tools, and Internet of Things connected devices articulate urban network infrastructures. Crashes that nowadays incur negligible injuries will considerably be avoided by advanced technological vehicles, and moderate-severity collisions are to be diminished to insignificant injury events.

   Urban big data analytics (Penmetsa et al., 2021; Yang et al., 2021) develops on virtual and augmented reality tools, sensing and navigation systems, cognitive automation technologies, and autonomous driving algorithms and behaviors. Self-driving cars may determine the inferring outcomes within short time, and consequently a big data-driven, prompt, and unfailing decision is typically achieved when the surrounding conditions transform significantly. Negative events such as fatal crashes or serious injuries negatively influence public perceptions as regards autonomous vehicle adoption.

   Deep learning and edge intelligence technologies, data tracking apps, virtual data modeling tools, remote sensor networks, and simulation optimization and motion planning algorithms (Stilgoe, 2021; Wang et al., 2022) reduce motor vehicle collision frequency. Urban sensing technologies integrate cloud and edge computing algorithms, virtual...