As the global population ages, addressing the transportation needs of older adults has become an urgent priority. In Canada, the demographic shift is particularly pronounced, with projections indicating that by 2030, approximately 22.8% of the population will be 65 or older. This trend challenges existing transportation infrastructure, and necessitates innovative solutions that consider the diverse mobility needs of seniors.
A promising approach is the development of Autonomous Mobility-on-Demand (AMoD) systems. These services provide on-demand, door-to-door service with autonomous vehicles, based on real-time user requests. It allows for flexible routing, scheduling, and vehicle allocation, enabling more efficient use of resources. This new mode of transportation could present a low-stress, affordable alternative to private cars for older adults, while the on-demand door-to-door aspect maintains the flexibility and independence provided by a private transport. However, accurately estimating these needs and understanding how they vary over time and space is crucial for the effective planning of such systems.
As the global population ages, addressing the transportation needs of older adults has become an urgent priority. In Canada, the demographic shift is particularly pronounced, with projections indicating that by 2030, approximately 22.8% of the population will be 65 or older. This trend challenges existing transportation infrastructure, and necessitates innovative solutions that consider the diverse mobility needs of seniors.
A promising approach is the development of Autonomous Mobility-on-Demand (AMoD) systems. These services provide on-demand, door-to-door service with autonomous vehicles, based on real-time user requests. It allows for flexible routing, scheduling, and vehicle allocation, enabling more efficient use of resources. This new mode of transportation could present a low-stress, affordable alternative to private cars for older adults, while the on-demand door-to-door aspect maintains the flexibility and independence provided by a private transport. However, accurately estimating these needs and understanding how they vary over time and space is crucial for the effective planning of such systems.
Modelling older adult mobility
Conventional approaches towards modelling mobility demand are limited by a lack of granularity at critical scales. Traditional four-step modelling approaches, which incrementally predict trip flows, mode uses, and route choices, work on the aggregate zonal scale. Complex inter-dependencies that might exist between demographic factors, trip generation, and other components (e.g. mode choice) are lost, and planning is unable to account for spatial and temporal heterogeneity in service utility.
An emerging alternative to traditional transport modelling approaches is agent-based modelling. This approach provides greater flexibility by capturing heterogeneity and dependency in and between contextual factors (e.g. demographics, health, household structures) and mobility preferences and behaviours.
Figure 1: approach for the proposed model
We designed an agent-based model (ABM) that focus on the heterogeneous mobility needs of older adults, and can serve as a basis to built up an AMoD service for older adults. In order to evaluate the older adults needs for AMoD, we focused on Winnipeg, the sixth most populous city in Canada. In 2021 Winnipeg had 749,607 inhabitants with 17 % of the population aged 65 or older. However, older adults’ spatial distribution is highly heterogeneous and in some areas, older adults can represent up to 70 % of the population.
Using primarily open data, we built an ABM of the elderly population’s daily mobility in Winnipeg on a standard weekday in 2022. The model is composed of a synthetic population with many socio-demographic characteristics, an activity-based model with individual activity chains and a model of the environment including the road network, public transit supply, buildings and facilities. The entire Winnipeg population is modelled – not only older adults – in order to reproduce interactions between individuals in households and in the environment.
Figure 2: synthetic population individual characteristics
During our work on this part of the ABM, we built an open-source synthetic population for the whole of Canada. Details of this data, and how to download it, can be found here: https://www.nature.com/articles/s41597-023-02030-4
Simulating older adult mobility with MATSim
Multi-Agent Transport Simulation (MATSim) is an open-source framework for transport simulations. MATSim was used to run simulations and calibrate the model to accurately reproduce the observed mobility. The model was validated by comparing simulated outcomes with real-world observations, to assess the reliability and accuracy of the model and ensure that the simulation adequately represents Winnipeg population and mobility.
Video screenshot of a simulation (speed x60). Each green arrow is a car on the road network. Visualisation with SimWrapper
A city level validation has been performed regarding Winnipeg global population distribution of sex, age and household size. The synthetic older adult population spatial distribution was validated through a comparison with the older adult population reported by the 2021 census. A quantitative validation of the general population mobility (trips purposes, modes, time) has been performed using the most recent travel survey for Winnipeg (2007). The mobility trends of older adults compared to the general population were then validated comparatively and using national studies. Key observations were accurately reproduced by the model and simulations:
- older adults travel less as they age and stay more at home;
- older adult mobility is heavily dependent on car use, with differences between ages and genders;
- older adults have temporal mobility patterns different from the global population with more trips in the morning until midday and before 4 pm.
Figure 3: simulated model share
Figure 4: simulated trips’ distribution by time of the day
Testing an AMoD service in simulation
A new AMoD service was introduced in simulation and some initial scenarios were tested with this new mobility option for older adults. The baseline scenario (no AMoD service) has been compared with scenarios where the new AMoD mode has been added for older adults. Older adult agents can thus switch from their initial mode to AMoD if this new mode is more convenient. The introduction of a new transportation service may also generate some induced demand by new users. Some scenarios thus explored situations where older adults with no car access would have the same mobility patterns as older adults with car access, but using the AmoD service. Simulation was used to test different fleet sizes and to explore the potential adoption of the AMoD service among Winnipeg older adults.
Figure 4: simulated trips’ distribution by time of the dayFigure 5: simulated modal share for elderly population, after introduction of 100, 250 and 500 AMoD shuttles
Implications for policy makers
This research may offer useful insights for urban planners and policymakers.
- If the older population value the AMoD service as much as they value cars, then the AMoD could become the main transportation mode for older adults. However, even in this case some will stick to the car, due to the additional waiting and detour time of the AMoD. We do not (yet) consider behaviours and social attitudes around latent attachment to cars.
- While increasing fleet size can initially improve overall accessibility, it is essential to consider the potential effects of induced demand.
- If an AMoD system provides a door-to-door service, the oldest groups might maintain a high level of mobility. While initial adopters might be men and non-licensed older adults, with a larger fleet, women and licensed older adults might switch to AMoD as well. Older women who do not have access to a car may use the AMoD service to make trips they could not make without a car.
- To avoid widening the mobility gender gap, it is essential to establish public policies and incentives that facilitate access to AMoD for disadvantaged populations, ensuring that pricing and access to service are tailored to their specific needs.
More details on this model, the results, and how it was constructed, can be found in our article in Transportation here: https://link.springer.com/article/10.1007/s11116-025-10590-0