The AV Vision Part 2: Accessibility, Sharing, Resilience and the Future of Public Transport
Key Takeaways
- This is the second blog in a series on a vision for autonomous vehicles (AVs). Last week's part 1 explored productivity, congestion, taxpayer funding and technology development; this week focuses on accessibility, public transport, shared mobility, resilience and road safety.
β
- AVs risk worsening accessibility through induced demand, congestion, and urban sprawl. However, they create an opportunity to repurpose land currently used for parking into housing, shops, better streetscapes, and improved walking and cycling infrastructure.
β
- Public transport remains essential in dense urban areas where space efficiency matters most, but AVs pose a real threat to its viability. Making buses driverless won't solve this; it may not significantly reduce costs and will likely reduce system resilience.
β
- Shared mobility will likely increase with AVs, but subsidies for private ownership and cultural attachments will limit uptake. We need deliberate policy intervention to maximise the shift toward sharing and capture its benefits.
β
- Resilience is a critical challenge. AVs struggle with adverse weather, power failures, and unexpected conditions. We must maintain alternatives (particularly human-operated public transport) and ensure multiple AV providers operate in each city.
β
- Road safety will improve as AVs become safer than human drivers. However, rushing deployment solely to reduce traffic deaths risks locking in car dependency and associated health problems for generations.
What Next?
Next weekβs blog will look at the social and environmental aspects of AVs and their implications for our AV vision.
Introduction
Autonomous vehicles promise to revolutionise urban transport, but the devil is in the details of how we deploy them.
Last week, in part 1 of this series, I examined the drier aspects of AVs: productivity, congestion, taxpayer funding, and technology development in urban environments. My conclusion was clear: AVs can deliver meaningful improvements over today's car-dominated transport systems, but only if we act now to shape their deployment.
This week, I'm exploring the implications for accessibility, public transport, shared versus private ownership, resilience, and road safety.
Accessibility
Accessibility is the ease with which people can reach desired activities, goods, or services (opportunities), such as jobs, shops, and schools. Itβs about the interaction between how we move and how we use land.
In the AV debate, it's tempting to focus exclusively on transport impacts and ignore land use. This would be a grave mistake. Like the cars that came before them, AVs will reshape our cities in profound ways.
The optimistic case is straightforward. AVs remove the burden of driving and, by eliminating driver costs, should be cheaper than today's taxis and rideshare services. Getting to jobs, schools, and shops becomes easier and more affordable. Accessibility improves.
The problem lies in the second-order effects. Cheap, easy mobility induces demand. More vehicles on the road means more congestion, which undermines the very accessibility gains AVs promise.
The second problem is spatial. If commuting becomes painless, people will rationally choose to live farther from work and amenities. This will drive urban sprawl and increase total vehicle kilometres travelled, worsening congestion.
But parking is where AVs present their greatest opportunity to improve accessibility. Shared AVs don't need to park near their destination; they can drop you off and leave. Privately owned AVs can park themselves in cheaper, more distant locations.
The implications are significant. Commuter car parks and city centre parking structures can be redeveloped as mixed-use neighbourhoods with housing, shops, offices, and services near public transport. Suburban shopping mall parking lots can be transformed into vibrant town centres with improved amenities and streetscapes. On-street parking can be converted to protected bike lanes and dedicated bus lanes.
Each of these changes increases accessibility by bringing destinations closer together and improving the quality of transport options to reach them.
In short, AVs could have significant positive and/or negative effects on accessibility. The trick will be to incorporate them into our transport systems and change our use of land in a way that minimises the drawbacks and maximises the opportunities.
Public Transport
The first question we need to ask is whether public transport has a future at all in a world of ubiquitous AVs.
This debate is already happening. In Ireland, commentators are questioning whether new metro systems make sense when cheap, convenient AVs might render them obsolete. The logic seems compelling: if AVs provide door-to-door service at comparable cost to public transport, why would anyone choose to wait at a bus stop or walk to a train station? Ridership collapses, the economics become unsustainable, and public transport withers away.
The people making this argument almost always have little understanding of transport systems. They forget two fundamental constraints. First, replacing public transport with individual vehicles, autonomous or not, creates massive additional demand on our roads. Second, and more critically, there is insufficient space in dense urban areas to accommodate all those vehicles. A train carrying 1,000 people occupies vastly less road and urban space than 1,000 cars, even if those cars are shared AVs making multiple trips.
In inner-city areas, we will still need to move large numbers of people efficiently through limited space. That means buses, trains, light rail, and metros.
But should public transport itself become autonomous? Some metro systems already are, at least partially, and many new lines are being built without drivers, reducing costs and improving reliability. The real debate is buses, the workhorses that carry the majority of public transport passengers.
Conceptually, autonomous buses offer clear benefits: lower operating costs, improved safety, and better reliability (no sick days or roster challenges). Currently, the transition is barely progressing beyond limited trials.
The challenges are substantial. New fleets are expensive. The technology isn't mature. And the business case is far from proven. I'm sceptical that buses will simply shed their driver and pocket the savings. Like autonomous cars, they'll need remote monitoring and intervention capability. During off-peak hours, when buses are quiet, safety concerns may necessitate on-board staff. Add union resistance and legitimate concerns about system resilience, and the economics appear much less attractive.
The real danger is that AVs undermine public transport. As ridership declines, service frequency is reduced, making public transport less attractive, which leads to more people using AVs in a downward spiral.
This is not inevitable, but it will happen unless we take deliberate action to prevent it. Our vision for AVs must account for the protection of public transport as an essential component of efficient, liveable cities.
Shared versus Private Mobility
The utopian AV vision always features shared mobility, fleets of robotaxis that arrive upon request, eliminating the need for private car ownership. Fewer total vehicles, reduced vehicle kilometres travelled, less parking, and better sustainability.
Will it actually happen?
There are reasons to be optimistic. Car share schemes have already demonstrated that many people will forgo private vehicles under the right conditions. AVs could significantly improve the offering: vehicles would come to your door rather than you walking to a parking bay, availability would increase through economies of scale as robotaxi and car-share services converge, and pricing should be competitive.
There's also a timing advantage. Fully autonomous private vehicles, in which you are not in control and don't need to monitor the system, may not be affordable for average consumers for many years. This creates a window during which shared AV services may be the only way most people can access autonomous mobility, potentially establishing enduring behavioural patterns.
But powerful forces work against sharing. We subsidise private car ownership: free or cheap parking, fuel and vehicle taxes that don't cover the actual costs of road infrastructure, and tax breaks for company cars. These subsidies create an uneven playing field, making ownership artificially attractive relative to sharing.
For many people, car ownership is also about more than transport; it's status, independence, convenience, and control. This emotional attachment is unlikely to change simply because of AVs.
Elon Musk has proposed a hybrid model where privately owned Teslas join a robotaxi network when their owners aren't using them. It's an unproven idea. Would owners really want strangers using their personal vehicle regularly? Would vehicles be available during peak-demand periods when owners themselves want to use them? From a transport systems perspective, this model is superior to pure private ownership but inferior to dedicated shared fleets.
The evidence suggests shared mobility will increase with AVs, but we won't reach the utopian vision of ubiquitous sharing. The default trajectory, shaped by existing subsidies, cultural preferences, and market forces, results in a mix that's better than today's but far from optimal.
If we want to maximise shared mobility and capture its benefits, we need to shape the incentives.
Resilience
When power failed across a third of San Francisco recently, Waymo's robotaxis struggled to cope. Some got stuck at intersections. The incident was a stark reminder: autonomous vehicles are fragile in ways human drivers are not.
One of the biggest challenges for AVs is coping with situations where they lack the data, technology, or smarts to handle them appropriately.
In addition to power failures, this includes fairly common weather conditions, such as storms and snow, as well as earthquakes. When these situations arise, today's AV operators have a simple response: pull the vehicles off the road. Better safe than sorry.
This works well now, when AVs are a novelty, and few people rely on them. It becomes a serious problem if AVs become the main mode of transport for a significant number of people. Imagine a snowstorm that shuts down an entire robotaxi fleet in a city where a quarter of residents have given up their cars.
Will AVs get better at handling adverse conditions? Almost certainly, given enough time. But this doesn't help if we create dependency before they are resilient enough.
Cybersecurity adds another layer of vulnerability. The nightmare scenario, someone hacking a fleet to run down pedestrians deliberately, is improbable but not impossible, as 9/11 demonstrated. A more plausible scenario is an attack that disables an entire fleet for days or weeks.
Then there's market risk. Recently, car-sharing operator Zipcar withdrew from the UK. Bike-share operators have also pulled out of markets. An AV operator could exit a market, go bankrupt after a catastrophic accident elsewhere, or be acquired by a competitor consolidating operations. If a city has become dependent on that service, the disruption could be severe.
Relying on a single shared AV provider creates dangerous fragility. A monopoly operator means no alternatives when things go wrong.
What does resilience require from our vision?
First, we need backup options. For the foreseeable future, people need access to transport modes that don't rely on AV technology. Buses are the obvious candidate, and there's a strong case for keeping them human-operated, or at least capable of human operation during crises, precisely because it provides resilience when autonomous systems fail. This is another reason why undermining public transport with AVs would be strategically foolish. Resilience also provides a strong rationale for investing in cycling infrastructure, as recent tube strikes in London have demonstrated.
Second, we need multiple AV providers in each city to increase resilience. If one fails or withdraws, alternatives exist.
Third, we need robust cybersecurity regulations and compliance regimes. But regulation alone isn't enough; we need contingency plans at the transport systems level for when (not if) failures occur. What happens when an AV fleet goes offline? How do we maintain mobility for essential workers, emergency services, and vulnerable populations?
Some cities may want to own and operate their own AV services, as they do with public transport. This provides more control but doesn't eliminate cybersecurity risks. As AV technology matures and becomes more widely available, and as barriers to entry fall over time, this may become more feasible.
The broader point is this: building a transport system that depends heavily on AVs means accepting new forms of fragility. We can manage these risks, but only if we plan accordingly.
Road Safety
On the surface, road safety seems like AVs' strongest selling point. I'm not going to wade into the debate about whether AVs are safer than human drivers right now. If you're interested, David Zipper has written an excellent analysis.
What matters more is the trajectory. AVs will continuously improve through software updates and machine learning from billions of kilometres of driving data. Human drivers won't get meaningfully better; we've been roughly the same for decades. Eventually, if we're not there already, AVs will be demonstrably safer than humans, even if the path includes setbacks and even if their failures look very different from human crashes.
This should make the Vision Zero goal of eliminating road deaths easier to achieve, but AVs alone won't get us there.
We need streets designed so that when crashes do occur, and they will, because perfection is impossible, the consequences are less severe. That means lower speed limits in areas with high pedestrian activity and residential neighbourhoods, low-traffic neighbourhoods, safe crossing opportunities along pedestrian desire lines, and protected infrastructure for cyclists.
Then there's the timing argument. Some commentators, usually people without transport backgrounds, argue that if AVs are safer than human drivers, we should deploy them as rapidly as possible to save lives.
This is one-dimensional thinking. Yes, when AVs are demonstrably safer, we might reduce traffic deaths more quickly with rapid AV deployment. But at what cost? Are we willing to lock in car dependency for generations and accept all the health (and many other) problems that it creates, just to reduce vehicle collision deaths slightly sooner?
Building transport systems that encourage walking, cycling, and public transport use delivers better overall health outcomes, even if it takes longer to eliminate traffic deaths specifically.
Rushing AV deployment solely to save lives from crashes, while ignoring the broader health and quality-of-life implications, isn't the right policy.
Conclusion
Like the first blog, this second blog in the AV vision series reveals a common thread: autonomous vehicles are not inherently good or bad for our cities; their impact depends entirely on how we choose to deploy them.
The accessibility gains from cheaper, easier mobility could be transformative, but only if we prevent induced demand and urban sprawl while seizing the opportunity to reclaim parking space for housing, shops, and better streetscapes.
Public transport remains essential in dense urban areas where space efficiency matters most. The shift toward shared mobility offers tremendous promise, but it won't happen without policy intervention to level the playing field against subsidised private ownership.
We need to think deeply about resilience. A transport system dependent on AVs could be inherently fragile, vulnerable to weather, power failures, cyberattacks, and corporate decisions. We will need to maintain alternatives, particularly human-operated buses that can function when autonomous systems cannot.
The road safety benefits are real, but rushing to deploy AVs as quickly as possible might reduce traffic deaths in the short term, but at the cost of other health problems, continued car dependency, and the problems it creates.
Next week, I'll examine the social and environmental dimensions of AVs and their implications for an AV vision.