Road use activity data: Cyclists, pedestrians and micromobility
This project designs, tests, and validates a pilot methodology for harvesting activity data of active road users – pedestrians, cyclists, and other personal mobility devices. It draws on experience in processing and analysing massive geospatial temporal data generated by users of smartphone devices.
A pilot algorithm will be developed to extract active trips from the SkedGo dataset and then validated and improve the estimation of activity data by active travellers in an iterative process. This process will use behavioural insights obtained from research on active travellers in Australia and secondary data including Household Travel Surveys, GPS tracking data and cycling surveys (counts, speed, and volume).
The project will create an R program to pilot the potential to collect ongoing active travel data (R is an opensource free software). The project report will assess the costs and benefits of continuing to update the data to inform road safety data collection for policy, program design and progress reporting.
It will have a particular focus on progress reporting for the National Road Safety Strategy 2021-2030 AND National Road Safety Action Plan 2023-25 by, for example, identifying the locations and share of high pedestrian activity areas, share of roads with separated cycleways in a particular speed limit (≥ 40 km/h) and/or environments such as urban areas.
Project background
For the past 50 years, transport modellers and planners have been using a variety of data collection methods to both describe and model travel demand and other travel choices. First and foremost are metropolitan Household Travel Surveys (HTSs), typically providing a snapshot of travel every 5-10 years, although done on a more regular basis in NSW and Victoria.
Supplementing these HTSs are a variety of counting programs and in-situ surveys capturing routine and project/location-specific information on system users, including vehicles, public transport users and cyclists/pedestrians. These approaches have known limitations, particularly around modal and spatial coverage, with the underrepresentation of active travellers (pedestrians, cyclists, e-cyclists, e-scooters and other micromobility modes) and regional/remote areas particularly acute issues.
New data collection techniques based on geolocation technologies of mobile/portable devices offer the potential for addressing these problems. The geolocation can take place either within the device’s own sensors and software or outside the devices using technologies developed by the system operators. Examples of the former include Bluetooth, GPS (e.g., Apple/Google), local Wi-Fi access points, and smartphone apps (e.g., Strava) while mobile network data, public transport smart cards, and vehicle telematics are examples of the latter.
All data collection sources have pros and cons, but those based on personal smartphones arguably have the most appeal for the problem at hand, because in addition to their virtual ubiquity within Australia (20.6 million Australians were using a smartphone in 2021), excellent spatial coverage, locational sensors and potential to interact directly with participants (if needed), they are likely to be carried by the majority of people whenever, however and wherever they move around. This makes them an ideal tool for collecting active travel information on a much larger scale than previously possible.
Smartphone-based solutions can be categorised as active or passive, the former requiring user interaction, typically through an app, the latter simply logging geolocational data in the background, which are used post-priori in the inference of travel information. Active methods have the potential to provide rich/high quality data at the level of the individual, by prompting users to provide/confirm inferred trip information as well as details on themselves.
However, they are hampered by practical issues, primarily requiring users to ‘opt-in’ to install an app and accentuate sample coverage/bias issues to those more comfortable with the technology.
By contrast, passive methods require little/no user interaction (possibly other than some agreement/consent that their information can be sequestered and used in the aggregate) and under an ‘opt-out’ model, can potentially expand sample size coverage to millions of Australians through, for example, subscribers to major network providers such as Optus, Telstra, and Vodafone.
The main challenge with passive methods, and to a lesser extent the active methods, remains how to infer population-wide travel/activity information using sensor-data alone while minimising (or preferably eliminating entirely) user interactions. If multiple-sensor data are available and reliable secondary sources (e.g., GIS networks), this is certainly possible, but under a truly passive solution we may only have the mobile network data to rely on.
Mobile phone tracking technology has already been used in the tourism and commercial sectors, providing mobile phone users with location-based services and it is increasingly used for transport modelling. The use of mobile phone data has been reviewed in transport modelling emphasising the paramount importance of understanding how the data will be used in designing and refining the process to generate activity data from mobile network operator and app data (see Figure 1).
The primary practical challenge for inferring population-wide travel/activity from app data is largely one of coverage, with the level of spatial coverage varying by number of app users. While this may not present major challenges in dense, urban areas with many app users, it becomes a significant challenge in rural and regional areas with fewer app users.
Pooling multiple days of data may help increase the spatial coverage. On the positive side, app data does not usually require rule-based algorithms for inferring the beginning and end location and time of trips and potentially modes and routes because these pieces of information are either provided by the app user as an input or generated by the app itself and confirmed by the app user via their choices. To this effect, active methods have an advantage over passive methods.
Project objectives
This project aims to design, test, and validate a pilot methodology for harvesting activity data on road use from active travellers (i.e., pedestrians, cyclists, and users of other micromobility modes such as e-scooters) and other personal mobility devices by drawing on experience in processing and analysing massive geospatial temporal data generated by mobile phone users across Australia.
Please note …
This page will be a living record of this project. As it matures, hits milestones, etc., we’ll continue to add information, links, images, interviews and more. Watch this space!
Discover more from iMOVE Australia Cooperative Research Centre | Transport R&D
Subscribe to get the latest posts sent to your email.
