Enabling the Future Internet for Smart Cities

Enabling the Future Internet for Smart Cities

InterSCity Mobility Research and Innovation

This page contains the mobility projects of InterSCity.

Digital Rails

Digital Rails is a system where autonomous vehicles would travel in platoons on exclusive lanes, achieving high efficiency through synchronization with traffic signals. The idea for Digital Rails (DR) began at Questtonó, a Brazilian innovation and design consultancy.


Digital Rails presents three main pillars:

  • Open data network: The city would provide a network that vehicles could access in order to obtain data about traffic signals, which would run in prefixed timing plans to allow uninterrupted progressions.
  • Exclusive lanes: On selected arterial roads, exclusive lanes from the current infrastructure would be assigned to the system. Vehicles using Digital Rails would have a custom design with narrower dimensions, so that a regular lane could fit two lanes of these custom-designed vehicles. One of them would be an expressway dedicated to platoons, and the other one would be used as extra space for maneuvers.
  • Vehicle platoons: Vehicles using Digital Rails would have some level of automation and organize themselves in platoons to maximize efficiency. The platoons would use the open data network to travel through the traffic signals progressions, minimizing travel time and number of stops. 

    The Figure bellow, extracted from Questtonó’s website, presents a concept-art for the project. The rationale for the project is to offer an alternative for urban transportation during the period while the autonomous vehicle fleet is growing, allowing AVs to share the roads with regular vehicles. For vehicles using the system, the travel time would be reduced, since there would be no stops due to traffic signals or traffic jams.

    Figure 1 - Digital Rails Concept

    Digital Rails Simulation

    We selected a region spanning a large portion of what is called the expanded downtown (centro expandido) of São Paulo, on which CET applies restriction of vehicle traffic during peak hours, based on their license plates. Figure 2 shows the region that we selected, represented on OpenStreetMap. The region consist in a box between latitudes -23.5054 and -23.6327, and between longitudes -46.7566 and -46.5707. It has a daily traffic magnitude of hundreds of thousands vehicles.

    We then proceeded to enumerate candidate arterial roads to host DR lanes inside the selected region. The main criteria used was the total traffic volume, as reported by CET. The rationale was that implementing DR lanes in the roads with the most volume should maximize the impact of the system. From the two most recent reports, we selected the following routes:

  • 23 de Maio Avenue
  • Pedro Álvares Cabral Avenue / Brasil Avenue
  • Marginal Pinheiros Avenue (between Jaguaré bridge and Eng. Ari Torres bridge)
  • Leste-Oeste Route
  • Cidade Jardim Avenue / Europa Avenue
  • Paulista Avenue
  • Eusébio Matoso Avenue / Rebouças AvenueThe map below shows the selected region with the routes highlighted in green. The next step was to generate realistic traffic input for simulation scenarios in the selected region. The starting point were the mobility trace of the city of São Paulo . We generated this dataset using the InterSCSimulator by running the traffic of the city of São Paulo for a whole day. From this trace, we isolated all vehicle that passed by the selected region at some point of the day. For each of these vehicles, we defined its trip starting and destinations points as the locations of its first and last appearance inside the region, respectively. This procedure yield an input file with more than 400k trips defined.

    Digital Rails Simulation


    As with the scenario with travels on Avenida Paulista, the travel time decreases when the ratio of vehicles able to use DR increases. It is noteworthy that a ratio of 25%, for example, does not mean that 25% of vehicles actually used DR. Instead, it indicates that 25% of vehicles were able to use DR and used the system whenever they went to a road with an assigned DR lane. We highlight the following results:

  • The average travel time is bigger than the average in the benchmark scenario when the ratio is 0. This is because the assignment of a lane for DR decreased the road capacity on the selected arterial ways.
  • With a ratio of 25% vehicles using DR, the average travel time were lower or very similar to the benchmark scenario on all distance quartiles.
  • For ratios greater than 50% of vehicles able to use DR, all average times were lower than the benchmark.
  • With 100% of vehicles able to use DR, the travel times were about 65% of the benchmark.
  • The evolution of travel times appears to be similar on all distance quartiles.We also analyze travel times considering only vehicles that are not able to use DR. Figure \ref{fig:dr-sp-out} shows the evolution of travel times for them, also divided in quartiles by travel distance. We highlight the following results:
  • With a ratio of 25% of vehicles able to use DR, the average travel time is equal or lower than in the benchmark scenario.
  • For ratios higher than 50%, the average travel time is smaller than in the benchmark scenario.
  • For 75% of vehicles able to use DR, the average travel time is between 67% and 79% of the benchmark scenario, depending on the considered quartile.
  • The evolution of travel times also appears to be similar on all distance quartiles.In conclusion, the simulated scenarios for DR in multiple arterial ways also present significant reductions in travel times, although less dramatic than the simpler scenarios on Avenida Paulista. This is not unexpected, since the paths of the simulated trips are not always within a road with a DR lane.

    Complete Capstone Project Text – Gustavo Covas