Facebook  Twitter  Linkedin  YouTube
Thursday , 22 February 2024

Improving public transport using data science

Dr Rajesh Krishnan, CEO, ITS Planners & Engineers

Providing better public transport for the travelling public requires a good understanding of travel demand so that buses can cater to the demand. However, bus networks in most cities have evolved over the years and many routes are operated due to legacy reasons, writes Dr Rajesh Krishnan, CEO, ITS Planners & Engineers

Various ITS systems have been deployed in our Smart Cities providing a number of datasets of raw, processed or modelled data. Cities could make use of these data sets and systems to estimate travel demand and improve public transport services in areas where demand exists. Some of the potential opportunities to do this better by making use of systems that are typically deployed in Smart Cities are given below.

Area Traffic Control System

ATCS are adaptive traffic control systems where the signal timings are adjusted automatically in response to changing traffic flows. Traffic detectors are installed as a part of ATCS systems in order to detect the traffic flows. Most Smart Cities have chosen to use modern traffic detectors for ATCS, such as video based detectors or multi-target tracking Doppler traffic radars, that provide reasonably accurate vehicle count data under mixed traffic conditions without lane discipline.

Most of the deployed ATCS systems also have the capability to be linked with micro- simulation software for transport modelling. Cities or ATCS vendors develop base models representing typical traffic conditions during different time periods of the day and days of the week and link these models with ATCS during deployment. The base models are automatically updated by the ATCS system using observed traffic count data. This essentially involves updating the travel demand patterns in the base model and estimating traffic flow patterns for the updated demand. The difference between modelled and measured traffic counts is reduced in the process resulting in more accurate demand than the base model. This means that where base models are linked with the ATCS, the system can be used to obtain transport demand over the course of a typical day. Alternatively, the count data can be downloaded and combined with the base model to estimate the demand manually when demand estimation functionality is not available in the deployed ATCS system. This takes a relatively modest amount of effort, and such analyses can be carried out using open source modelling software such as SUMO without incurring a high cost.

Automatic Traffic Counters and Classifiers

Some Smart Cities have installed Automatic Traffic Counters and Classifiers. ATCCs are high accuracy traffic detectors that are typically deployed mid-link on major arterials within the city and at major entry/exit points to the city on the outskirts. Data from ATCCs can be used in a similar fashion to count data from ATCS to estimate time dependent spatial travel demand patterns.

Automatic Number Plate Recognition Cameras

ANPR cameras record the registration number of passing vehicles along with the timestamp. A network of ANPR cameras are installed in a number of Smart Cities, some times as a part of enforcement systems and stand alone in some cases. Even though ANPR cameras do not capture all the vehicles, this data source provides a sample of origin-destination movements of vehicles, and sometimes information about the route taken depending upon the number and locations of ANPR cameras. As most Smart Cities also have transport models from the ATCS system, origin- destination data from ANPR can be used to update the demand in the base models when the sample size of ANPR data is large enough. This is an offline activity involving a modelling software outlined above. Updating the demand in the base model will result in more accurate estimates of origin-destination travel demand .

Data from mobile telecom systems

Origin-destination demand can be estimated more accurately compared to the above methods when cities have access to mobile telecom data. Mobile telecom systems record events such as calls, SMS and data exchanges, the IMEI number of the mobile device used and the mobile tower a mobile device is connected to during the event along with the timestamp. A pseudonymised version of this dataset from any major telecom operator provides a rich sample of the movement of the population in space and time when combined with the location data of mobile towers. The sample travel patterns from telecom operators can be combined with traffic counts with the aid of a modelling software to estimate overall travel demand in cities. This technology has been available in the country for many years now and is accepted in practice in a number of countries outside India.

Using demand data for public transport

The above methods present feasible options for cities to model and understand travel demand in cities in a temporally granular fashion. This will help city authorities understand the requirement for mobility geographically during the course of a day. Cities can thus design their public transport routes and time tables based on a solid understanding of demand data. Such a data driven design and operation of the public transport system will give the public a feasible alternative to private transport, potentially resulting in mode shift and reduced congestion.

Share with: