Smart Transportation: From Technology and Machine to Infrastructure and Human

Written by Dr. Shen Wang

Smart transportation sessions are one of the key activities of the annual IEEE International Smart Cities Conference (ISC2). This year, smart transportation sessions received the highest number of submissions among all technical sessions. Our sessions cover a wide range of topics from urban transportation infrastructure planning, to autonomous vehicle related technologies, and even include airport bottleneck analysis. There are two main trends clearly shown in our sessions this year. Firstly, the research focus of smart transportation is getting more integrated with the infrastructure design of the city. While big data and artificial intelligence (AI) are still of interest particularly in traffic prediction and autonomous driving, we have seen a rising number of research on building testbeds for demonstrating the design, or using a realistic dataset to verify the replanning of important transportation infrastructure. Secondly, the focus of smart transportation is shifting towards “human-centric”, rather than "machine-centric". Examples include a prototype system detecting driver's distraction and new communication protocols for ensuring the safety of vulnerable road users.


Trend 1: From Technologies to Infrastructure

Many papers in the previous years’ smart transportation sessions are about information and communications technologies (ICT) themselves. Typical examples include using AI such as deep neuron networks (DNN) to predict urban road traffic, and using computer vision technologies to detect the road objects for autonomous driving. The focus of such research is mainly on increasing the accuracy of traffic prediction and object detection within strict time frames. Similar research can also be found this year: Sabour et al. [12] proposed a DNN-based algorithm using a Siamese network dubbed “DeepFlow” to predict the traffic events, which is the abnormal traffic flow with a verified higher performance based on a widely used simulator platform. A study in [1] uses a similar method for predicting cyclist trajectory with multi-modal data sources. Another preliminary study [15] proposes an abstraction of data collection for autonomous driving, so that the performance would not be heavily biased on the physical characteristics of various sensors for collecting data. A similar study [6] also focuses on autonomous driving by proposing an IoT system to generate dynamic maps for safer vehicle path planning.

However, a clear trend found this year is that more work starts focusing on the interactions with or planning transportation infrastructure, in which ICT stepping down from the main role to an assisting role to facilitate the infrastructure planning. Ohsugi et al. [4] define a practical problem based on the traditional travelling salesman problem to decrease the re-deliver ratio due to the absence of the receiver at home. They also proposed a solution to this problem, which is powered by the data gathered from the smart meter infrastructure to predicted if the targeted users are at home. For the campus parking infrastructure planning,  Vechione et al. [5] conducted a field study using very few data collection points over the campus to calibrate a traffic demand prediction model. Kampyli et al. [9] proposed a new idea using crowdsourcing for collecting the reliable parking demand and supply information, so that the urban parking infrastructure can be well utilised. Ye et al. [13] improve the performance of the short-term forecasting so that the utilization rates of electric vehicle charging stations will be more balanced. S. Agrawal and G. Elger [16] implemented a test field for off-loading the massive computation and data storage from vehicles to the road side unit (RSU) infrastructure. This test field enables further improvement for the environment perception of autonomous vehicles. Particularly focused on the urban junction [17] Hetzel et al. developed a networked sensor system that is deployed in the city of Aschaffenburg, Germany, for ultra-high resolution real-time traffic collections.

 

Trend 2: From “Machine-Centric” to “Human-Centric”

Another trend that can be seen clearly is a shifting of research focus from “machine-centric” to “human-centric”. The general examples of “machine-centric” research include the sensor and path planning technologies of connected autonomous vehicles and DNN based traffic prediction. Rather than improving the machine performance only, the increasing growth of “human-centric” smart transportation is widely seen in this year’s session with more realistic dataset support and more practical research questions posed for human benefits.

Gonçalves and Paiva [2] proposed a system developed using Google Cloud Vision for visually impaired people. Ebata et al. [3] applied users’ dissatisfaction to bridge the so-called “short-distance transportation gap” with supply-and-demand mediation type service. They achieved a 40% reduction in walking distances according to a simulation study on the Tokyo Bay area. Lago et al. [7] developed a proof-of-concept that can use a camera to detect if the driver is distracted while driving, according to the threshold set by abnormal head moving angle. Banjade et al. [8]  from Intel devised a new communication protocol that can ensure better safety for vulnerable road users. Lai et al. [10] applied a realistic dataset to study the bottleneck of the airport at JFK and made suggestions to improve the service so that the waiting time for passengers will be significantly shortened. Li et al. [11] utilise mixture density networks and stochastic programming for improving the performance of a ride-hailing system to better guide the driver. Benchimol et al.  [14] developed a web-based application to enhance the travellers’ experience in Paris. Using machine learning models powered by the train passengers’ load dataset, the trips recommended by this application supports cover multi-modal transport.

 

Conclusions

This year’s smart transportation sessions are witnessing progress in well-known research domains such as autonomous driving and traffic prediction. However, there are two trends that still can be clearly seen in comparison to previous years’ sessions. “From technologies to infrastructure”: more work is attempting to leverage ICT for better infrastructure planning/re-planning, rather than improving the efficiency of the ICT themselves. “From machine-centric to human-centric” further shifting the research from big data and AI to the potential benefit they can have on a human. Overall, we see these are two very positive trends, which are bringing smart transportations more integrated with the whole smart cities, especially focusing on the benefits on the end-users (each citizen). We look forward to seeing the continuous growth in these two directions, maybe include more exciting areas that we cannot even think of now. See you all next year!

 

 

 

References:

  1. S. Zernetsch, O. Trupp, V. Kress, K. Doll, and B. Sick, " Cyclist Trajectory Forecasts by Incorporation of Multi-View Video Information," 2021 IEEE International Smart Cities Conference (ISC2)
  2. J. Gonçalves and S. Paiva, "Inclusive Mobility Solution for Visually Impaired People using Google Cloud Vision," 2021 IEEE International Smart Cities Conference (ISC2)
  3. T. Ebata, M. Imamura, and K. Suzuki, "Using User Dissatisfaction to Bridge Transportation Gap with Supply-and-demand Mediation-type Service," 2021 IEEE International Smart Cities Conference (ISC2)
  4. S. Ohsugi, S. Negishi, K. Okada, H. Yoshii, K. Tanaka, and N. Koshizuka, "Traveling Salesman Problem on Smart Meter Infrastructure," 2021 IEEE International Smart Cities Conference (ISC2)
  5. M. Vechione, S. Paudel, and O. Gurbuz, "Trip Distribution Patterns on a University Campus: A Smarter Travel Demand Forecasting Approach," 2021 IEEE International Smart Cities Conference (ISC2)
  6. M. Moussa and L. Alazzawi, "IoT-Based Dynamic Map Attributes for Connected and Autonomous Vehicles," 2021 IEEE International Smart Cities Conference (ISC2)
  7. T. Lago, E. González, and M. Campista, " Towards a Real-time System based on Regression Model to Evaluate Driver's Attention," 2021 IEEE International Smart Cities Conference (ISC2)
  8. V. Banjade, S. Jha, K. Sivanesan, L. Baltar, S. Sehra, and S. Tan, "Vulnerable Road Users Safety in Infrastructure Assisted Intelligent Transportation System," 2021 IEEE International Smart Cities Conference (ISC2)
  9. A. Kampyli, S. Kontogiannis, D. Kypriadis, and C. Zaroliagis, " Incentivizing Truthfulness in Crowdsourced Parking Ecosystems," 2021 IEEE International Smart Cities Conference (ISC2)
  10. J. Lai, L. Che, and R. Kashef, "Bottleneck Analysis in JFK Using Discrete Event Simulation: An Airport Queuing Model," 2021 IEEE International Smart Cities Conference (ISC2)
  11. X. Li, J. Gao, C. Wang, X. Huang, and Y. Nie, "Driver guidance and rebalancing in ride-hailing systems through mixture density networks and stochastic programming," 2021 IEEE International Smart Cities Conference (ISC2)
  12. S. Sabour, S. Rao, and M. Ghaderi, "DeepFlow: Abnormal Traffic Flow Detection using Siamese Networks," 2021 IEEE International Smart Cities Conference (ISC2)
  13. Z. Ye, R. Wei, and N. Yu, "Short-term Forecasting for Utilization Rates of Electric Vehicle Charging Stations," 2021 IEEE International Smart Cities Conference (ISC2)
  14. P. Benchimol, A. Amrani, and M. Khouadjia, "A Multi-Criteria Multi-Modal Predictive Trip Planner: Application on Paris Metropolitan Network," 2021 IEEE International Smart Cities Conference (ISC2) [Short]
  15. H. Reichert, L. Lang, K. Rösch, D. Bogdoll, K. Doll, B. Sick, H. Reuss, C. Stiller, and J. Zöllner, "Towards Sensor Data Abstraction of Autonomous Vehicle Perception Systems," 2021 IEEE International Smart Cities Conference (ISC2) [Short]
  16. S. Agrawal and G. Elger, "Concept of Infrastructure Based Environment Perception for IN2Lab test field for Automated Driving," 2021 IEEE International Smart Cities Conference (ISC2) [Short]
  17. M. Hetzel, H. Reichert, K. Doll, and B. Sick, " Smart Infrastructure: A Research Junction," 2021 IEEE International Smart Cities Conference (ISC2) [Short]

 

 

 

This article was edited by Aris Gkoulalas-Divanis

To view all articles in this issue, please go to December 2021 eNewsletter. For a downloadable copy, please visit the IEEE Smart Cities Resource Center.

shen picture small
Shen Wang is currently an Assistant Professor with the School of Computer Science, University College Dublin, Ireland. He received the M.Eng. degree from Wuhan University, China, and the Ph.D. degree from the Dublin City University, Ireland. Dr. Wang is a member of the IEEE and a reviewer of its major conferences and journals in intelligent transportation systems. His research interests include trajectory data mining and processing, connected autonomous vehicles, and explainable artificial intelligence.