Chapter 08

What To Do Now

A Sequenced Action Plan for Cities, Utilities, and Investors

8.1The governance window

The regulatory landscape for connected autonomous mobility is in active formation. The UK’s Automated Vehicles Act received Royal Assent in 2024 and is expected to enable fully driverless commercial deployment as its remaining operational regulations complete.^EV58 Meanwhile, the first autonomous services entered under existing private-hire licensing in August 2026, when Transport for London granted fifteen licenses for autonomous services with safety drivers aboard. The EU’s AI Act classifies autonomous driving systems as high-risk, with requirements including data governance, human oversight, and post-market surveillance, but applies them through vehicle type-approval law rather than directly. Under a provisional agreement reached in May 2026, these take effect for autonomous vehicles on 2 August 2028^EV98, a year later than first legislated. In the United States, NHTSA has signaled four new rules updating federal safety standards for vehicles with automated driving systems, while federal framework legislation remains in committee.^EV53 The UN Economic Commission for Europe’s (UNECE) global Level 4 regulation is expected during 2026.^EV52 Across all major jurisdictions, the formal governance framework for autonomous mobility will be in more settled form by the end of 2026 than it is today.

The US legislation referenced above – the SELF DRIVE Act of 2026 – bears closer attention. Passed by the House Energy and Commerce Committee on a 12–11 vote in 2026, the bill would pre-empt state and local rules that “prohibit or partially prohibit” autonomous vehicle deployment by operators with a federally recognized safety case.^EV64 Such a rule would override much of Texas SB 2807’s^EV3458 enforcement authority and a wide range of city-level operating frameworks. The principle behind federal pre-emption is logical for vehicle standards and automated driving system (ADS) performance, where city-by-city variation would be unworkable. It does not hold for the operating conditions of fleet activity within a city’s jurisdiction – matters of curb governance, service obligations, equity coverage, fiscal model. Here, municipal authority is critical to legitimate deployment. Cities that engage with the federal legislative process while it remains in formation are the ones most likely to see that distinction preserved in the final framework.

Comprehensive frameworks are emerging across all major jurisdictions, even where specific rules remain unpublished. The structure around urban autonomy is hardening. The governance window, in other words, is closing, decision by decision. Each closure has an identifiable threshold beyond which reversal becomes materially more expensive. All of them are visible in advance to a city that knows to look. Cities that engage with the regulatory process while it is in formation can secure data-sharing provisions, licensing frameworks, and city-level oversight mechanisms that reflect local governance needs. Cities that engaged early with TNC governance – New York, London, Chicago – achieved better regulatory outcomes than cities that waited for federal or state frameworks to arrive.

This chapter translates the Governability Stack, the Institutional ODD, and the five levers of value at risk into a sequenced action plan organized by timeline and stakeholder. Each action is deployable under existing institutional authority, at manageable cost, with precedents in analogous governance contexts. They are the minimum viable response to a structural governance challenge whose timeline is now determined by regulatory calendars rather than by technology readiness.

8.2No-regrets actions: 90 days

The first 90 days establish the foundation. These are decisions a city can make immediately, with no dependency on national regulatory outcomes, at low cost, and with no risk of regret regardless of how CAM deployment unfolds. They are no-regrets in the strict sense: even if autonomous vehicles take longer to scale than current trajectories suggest, these actions build institutional capacity that is valuable for managing existing and emerging mobility systems. No-regrets is used in its climate-adaptation sense; robust across scenarios. Senior leaders’ convening capacity is often a scarce resource, and using it for CAM is costly.

Score your Institutional ODD across all four domains

Using the framework set out in this paper, assess your city’s current institutional capacity across Flows, Space, Power, and Data. The scoring is simple: for each domain, does your existing institutional framework allow you to set conditions, enforce compliance, collect relevant data, and adjust instruments in response to evidence? In most cities the answer for at least two of the four domains is no. That is the starting point. Document the assessment. Share it internally across departments. It becomes the baseline against which all subsequent institutional reform is measured, and the evidence base for the cross-agency governance case that must be made to elected leadership. How the assessment is conducted matters: a candid score of a city’s own ungovernability is discoverable, and career staff asked to author it unprotected will score it defensively. Conduct it the way cyber-maturity and clinical-governance reviews are conducted – through an external assessor or under privilege, with aggregate results shared and detail held.

Introduced in 2.3 Technical ODD and the Institutional ODD

Audit existing authority against the instrument menu

Before designing new governance frameworks, map what you already have. Most cities possess regulatory instruments for CAM governance that have not been activated: curb allocation frameworks, commercial licensing powers, data-sharing provisions in existing transport agreements, design standard requirements in planning policy. The audit question is “what instruments do we already hold that we have not deployed, and what cross-agency dependencies stand between them and use?” The answer is usually that much of the toolkit exists, fragmented across agencies – which is less reassuring than it sounds: integration is the hard institutional problem, and much of what is missing cannot be invented locally without state enabling legislation.

Introduced in 3.6 The Governability Stack

Initiate the minimum viable operating picture

Begin collecting structured data from mobility operators already active in your jurisdiction – scooter operators, taxi fleets, bike share programs, demand-responsive transport services.^EV3280 No legislation is needed; a proof-of-concept partnership establishes three things at once: the precedent that agencies and operators can exchange structured data without multi-month negotiation cycles, the technical infrastructure for data ingestion and processing that will later be required for CAM data, and the habit of treating operational data as a governance resource. By the time autonomous fleets are active in your city, the operating picture should already exist and be running.

Introduced in 5.3 The minimum viable operating picture

Establish the cross-agency governance wrapper

Autonomous mobility does not fit into existing departmental silos. It touches parking management, transport strategy, highway licensing, emergency planning, utilities coordination, planning enforcement, and in some jurisdictions insurance liability. Create a formal governance structure – a cross-agency steering committee or autonomous mobility authority with explicit decision rights – that owns the Institutional ODD and the city’s adaptive governance capacity. Give it quarterly reporting obligations to elected leadership, budget authority over pilot programs, and a mandate that is explicitly cross-domain, not a single department’s brief.

Almost every precedent has a forcing function, such as a statutory mandate, a mayoral priority, or a live emergency. A convening without one is hollow. The realistic options are a mayoral executive order or a council resolution that attaches budget; cities that wait for the third option, an incident, tend to get a convening anyway.

Introduced in 6.2 Institutional reforms: the operating authority uplift

Engage operators on data-sharing expectations

Before regulation mandates it, communicate clearly to the operators active in or approaching your market what data requirements will be attached to any future deployment authorization. Frame it as the operating condition for sustained, legitimate deployment. Operators who understand the city’s governance expectations in advance can design their data systems accordingly, reducing the cost of compliance and eliminating the adversarial dynamic that makes negotiating data-sharing requirements after deployment so politically difficult.

Introduced in 6.2 Institutional reforms: the operating authority uplift

8.3Building capacity: 12 months

Having established the institutional foundations, the 12-month horizon focuses on translating frameworks into operational systems. These actions require meaningful institutional investment but remain low risk in the sense that all of them build capabilities that are valuable across a wide range of deployment scenarios. They also begin to create the external commitments – to elected leadership, to operators, to the public – that transform institutional intention into institutional practice.

Design the licensing framework

A CAM license should specify, at minimum: vehicle class and autonomy level covered by the authorization, geofence limits for initial deployment, maximum permitted fleet size with adjustment mechanisms, mandatory data-sharing conditions including format, frequency, and retention standards, service obligations covering geographic coverage, accessibility requirements, and equity commitments, safety reporting thresholds and escalation procedures, financial securities protecting the city’s costs in the event of operator withdrawal, and fee structures providing for full cost recovery of city oversight. The license is a governance contract. Write it that way, using the institutional instrument menu from Chapter 6 as the design guide and the TfL micromobility concession and New York FHV regime^EV33 as operational templates. And take it to public consultation before the conditions settle. The paper’s own logic requires this: Chapter 7 sets democratic endorsement as the standard for a reversible street pilot, and a license schedule that will fix fiscal model, coverage and workforce terms for a decade cannot carry a lower legitimation requirement than a row of planters.

Introduced in 6.2 Institutional reforms: the operating authority uplift

Pilot reversible street and curb interventions

The spatial system is the slowest to change and therefore the first to create lasting patterns. Test modular curb allocations for fleet pickup and drop-off, micro-terminal designs at high-demand locations, dynamic pricing on a small number of arterial curb segments, and managed loading zones with time-of-day allocation. Apply the governance lessons from the LTN experience described in Chapter 7 and budget for the governance, not only the planters. Budget explicitly for governance infrastructure, not only for physical materials. Consider measurement systems, community engagement, coordination meetings. Physical success without governance success will quickly be removed.

Introduced in 7.2 Streets and rights-of-way

Commission the fiscal transition plan

Parking revenue will decline as fleet vehicles replace private car trips and reduce parking demand. Enforcement income will fall as automated compliance detection changes the enforcement landscape. Transit farebox revenue faces pressure if autonomous services compete with transit in ways that draw off financially sustainable ridership without serving the trips that are most dependent on public subsidy. The fiscal transition plan should model all five levers of value at risk from Chapter 4 for your city’s specific revenue profile, identify replacement mechanisms for each revenue stream at risk, and sequence the introduction of those mechanisms to prevent the budget gap from appearing before the replacements are operational. Elected officials need to see this plan before the crisis emerges, not after – because the political authority to act is much easier to secure when the fiscal risk is prospective rather than immediate. One asymmetry should be put to elected leaders directly: this is the plan whose costs fall inside the current electoral term and whose benefits largely accrue beyond it. No sequencing removes that; it is the one unavoidably cross-term commitment in the program, and it should be presented to elected leaders as exactly that.

Introduced in 4.2 The five levers of value creation

Coordinate energy and depot planning

Utilities in most jurisdictions are not currently engaged in joint planning with transport authorities on autonomous mobility deployment. This must change, and the 12-month horizon is the appropriate moment to initiate that change while depot siting decisions have not yet been made. The joint planning process should map grid hosting capacity by geography, identify the corridors where depot density is technically feasible without requiring major network reinforcement, establish managed charging standards for any new depot development, and create a formal data-sharing arrangement between transport and energy authorities on fleet charging load forecasting. The city that treats grid capacity as a governance lever rather than an external constraint will be able to guide operator depot siting towards locations that serve both transport and energy objectives.

Introduced in 7.4 Depots and charging infrastructure

8.4Institutional maturity: 3–5 years

The three-to-five-year horizon is where institutional capacity becomes embedded and self-reinforcing. The actions in this phase build on the foundations and frameworks established in the first two phases, converting one-off governance decisions into operational systems and political commitments into institutional habits. They also represent the point at which the city’s governance framework must demonstrate its value not only to internal stakeholders but to operators, investors, and the public – the constituencies whose continued engagement is required for the adaptive city to function as more than an aspiration.

Deploy the full operating picture

By year three, the governance information system should be operational across all of the data capabilities set out in Chapter 5: mobility and curb operations, energy and grid performance, land and real estate transition, fiscal and pricing outcomes, equity and access patterns, and compliance and audit performance. It is a permanent governance system, operated and maintained by the city as core infrastructure, used to inform quarterly policy reviews and annual licensing decisions. The operating picture also enables the simulation capability described in Chapter 5: the ability to model deployment scenarios before making licensing, infrastructure, and pricing decisions, and to test instrument calibrations against modeled outcomes before committing to them in the live operational environment.

Introduced in 5.4 From operating picture to simulation

Implement the economic toolbox at network scale

Having tested dynamic curb pricing on individual corridors and managed loading zones in defined areas, the 3-to-5-year horizon should see these instruments integrated into a network-wide pricing framework. Congestion charges, curb rent, fleet charging tariffs, and the value capture mechanisms described in Chapter 7 should operate as an interconnected system in which pricing signals reflect both marginal infrastructure cost and explicit policy objectives: modal shift, depot dispersal, equity of service distribution. The sequencing principle from Chapter 6 applies: start with the instruments that carry the lowest political risk and strongest empirical foundation, use operating picture evidence to demonstrate their effectiveness, and use that evidence base to build the political case for the subsequent instruments in the sequence.

Introduced in 6.3 The economic toolbox

Embed conversion-ready design standards

By year five, every important land use decision – new parking construction, surface lot development, logistics warehouse planning consent – should be evaluated against the conversion-ready specifications described in Chapter 7. Cities in which these standards are embedded in planning policy before major development cycles will have a built environment that can adapt to changing parking economics without requiring crisis-driven retrofit programs.

Introduced in 7.5 Buildings and parcels

Build institutional memory through the recursive learning cycle

Governance that does not learn from evidence degrades. Establish a formal process – annual or biennial – for reviewing actual deployment outcomes against the five-domain value-at-risk projections. Which forecasts proved accurate? Which assumptions broke down when exposed to operational reality? How did operator behavior differ from the models used to calibrate licensing conditions? Embed the answers to these questions into updated governance frameworks, and document the reasoning so that institutional knowledge persists beyond individual tenure. The city that learns faster than its operators and adapts its governance instruments in response to evidence will maintain regulatory effectiveness as deployment scales.

Introduced in 1.3 Unknowability and the case for option value

8.5What each actor owes

The action plan does not fall equally across stakeholders. The governance challenge of connected autonomous mobility is cross-sector, requiring different institutions to make different contributions at different points in the sequence. What follows is the minimum each actor must contribute for the adaptive governance system to function.

City leadership and multi-agency executives. You own the Institutional ODD assessment, the cross-agency governance wrapper, and the political mandate for fiscal transition planning. These responsibilities are not delegable to individual departments or to specialist transport units. They require executive commitment, budget authority, and quarterly reporting to elected leadership. The cities that have failed at TNC and micromobility governance have done so because leadership treated the challenge as a specialist transport problem amenable to standard departmental decision-making. It is a cross-domain challenge that requires the same executive authority and cross-agency coordination that major urban crises and major infrastructure programs command. The 90-day actions described above are your starting point. They cost little and create the foundations for everything that follows.

Utilities and energy regulators. Your role in CAM governance is larger than current institutional arrangements acknowledge. Grid hosting capacity is the binding constraint on depot density, charging infrastructure scale, and therefore on the physical feasibility of fleet operations in specific locations. You hold the data that transport planners need to make rational siting decisions, and you hold the technical expertise to design managed charging systems that convert fleet charging from a grid liability into a grid asset. The 90-day action for utilities is to share hosting capacity maps with transport authorities and to establish joint planning protocols that make energy infrastructure a first-order consideration in licensing and siting decisions rather than an afterthought.^EV96 Where the energy failure mode is one of coordination – nobody in the city knew a depot was coming until the application landed – it is preventable at low cost. Where the constraint is real scarcity on a congested feeder, it is an allocation problem with real losers. Which of the two dominates is an open empirical question. We do not know the answer, and a utility that claims to should be asked for its hosting-capacity data.

Real estate owners and investors. Audit your parking portfolios for conversion feasibility. A surface parking lot in a dense, well-connected urban area with adequate electrical supply is, in the medium term, a logistics asset, a charging hub, or flexible light-industrial space, depending on its size, location, and grid connection. Know what your assets can become before the market prices the transition. Position early in corridors where the city’s regulatory and fiscal framework signals sustained public commitment to CAM deployment. The governed dividend accrues to assets whose owners built optionality from the start. It does not accrue retrospectively to standard-specification assets that require expensive retrofit to participate.

Operators and logistics companies. The operators who treat city governance as a constraint to minimize will consistently achieve worse commercial outcomes than the operators who treat it as a framework to engage. Uber’s confrontational approach won markets and then spent a decade defending them – in court in London and Barcelona, and at the ballot box in Austin, where it lost a 2016 vote, withdrew for a year, and returned only when the state overrode the city. Operators entering governed markets now can skip that decade. The operators who move first to share data before mandates require it, to support licensing frameworks that create visibility, and to embed themselves in city planning processes rather than circumventing them, are the operators who shape regulatory precedent in their markets. Operators who understand this will find that the cities are not adversaries but the institutional partners who make large-scale deployment possible.

8.6What this paper does not yet contain

A paper that stakes its credibility on a visible evidence base owes the reader an equally visible account of its gaps. Six pieces of the toolkit are named in this edition and remain undesigned – we see them as commitments to a research agenda rather than caveats.

The first is the regulatory-intervention instrument for the Power domain: Chapter 6’s menu contains no entry for the lever that actually governs investor-owned-utility behavior. This could include rate cases, connection-policy consultations, hosting-capacity disclosure requirements. Designing one requires jurisdiction-specific work on standing, timing and evidence.

The second is the per-vehicle-mile component of the fee structure, tracking empty miles alongside passenger miles. This depends on a data feed or audit regime for in-boundary mileage.

The third is a governance mechanism for the city as data holder. A data governance standard would specify retention, purpose limitations, secondary disclosures, and independent oversight, matching the rigor expected of operators.

The fourth is a worked conversion case for Chapter 7’s design premium. Using a real-world building, the case would develop a named end use, a stated rent, an assumed horizon and a break-even model. With multiple cases, we could derive patterns and lessons.

The fifth is the detailed design of the workforce transition mechanism named as a license condition in Chapter 4. And running under all five is the analytical layer this paper states as an open question wherever it arises.

Over the coming months, we will see whether the rate of autonomous vehicle deployment is limited by fleet supply or demand density. The answer redistributes emphasis between this paper’s Power and Space domains. The Lab will be tracking fleet deployment and ridership patterns, and publishing the results of analysis as frequently as they arrive. This paper is built on a living evidence base that will be continually updated and tagged to themes in the paper.

8.7The adaptive city

A connection agreement negotiated bilaterally between an autonomous fleet operator and a grid utility specifies the ideal charging time block, beginning at 1am. As the clock strikes, an electron leaves the regional substation. It passes through a feeder circuit that was recently upgraded to accommodate the peak demand associated with fleet electrification. The electron travels to a part of the city where land is cheap, and arrives at a parcel zoned for light industry. Here, it enters a charging depot that is owned by a private garage owner and financed together with the city’s development agency – a long-term bet on public infrastructure. The electron moves through a Level 3 charging station that is owned and managed by an autonomous fleet operator, and finds its way into a vehicle battery.

By the time morning rush hour begins, the vehicle is sufficiently charged. It pulls out of the depot and begins traveling according to a routing policy that the operator optimized within constraints set by the city’s transportation department. The vehicle is called by a commuter. It checks city-wide congestion patterns, and takes a residential street as a through route to ease traffic on a main road. When it stops at a curb, the fleet operator begins paying for curb access time – a dynamic rate that floats with demand. The passenger enters, and the car departs. Within a few blocks, electrons come flooding back into the battery through its regenerative braking system – the vehicle has slowed to accommodate a wheelchair user who is moving slowly through a crosswalk. The operator service contract specifies preemption for vulnerable road users.

At the end of its shift the car is plugged back into the charging station at the depot. The grid is at peak usage, so the electron flows back onto the feeder it started from, routed, eventually, to an apartment that needs afternoon air conditioning. It is priced at a tariff rate the energy regulator approved, metered, and documented in data the city is entitled to read.

One electron crosses all four domains of the Governability Stack and at least six different institutions. The protocols that make this system work must be written by the distribution network operator, the transport authority, the planning authority, real estate owners, and fleet operators cooperatively. Until they do, not one of them can see its whole path.

The purpose of the Governability Stack is to demonstrate interdependence. No domain can be governed in isolation, and no instrument operates cleanly within a single domain. Routing decisions in the Flows domain determine curb demand in the Space domain. Depot siting decisions in the Space domain determine grid load profiles in the Power domain. Charging data in the Power domain enables peak demand management through economic instruments in the same domain. Compliance monitoring in the Data domain enforces licensing conditions in the Flows domain and curb pricing in the Space domain. A congestion charge is simultaneously a Flows intervention, a Data prerequisite (you need to be able to see what is happening to price it), and a Power implication (lower fleet VMT means lower aggregate charging demand). Every governance decision in the matrix is simultaneously a decision about multiple cells.

A city that is truly adaptive does not need to attempt predictions about the future of autonomous mobility. The governance toolkit described in this paper – the operating picture, the licensing framework, the economic instruments, the spatial design standards, the cross-agency coordination structure – is designed to work across the full range of deployment scenarios that currently appear plausible. It works for dense robotaxi networks in major cities and for autonomous freight operations in mid-sized logistics hubs. It works for mixed human-autonomous fleets on strategic corridors and for single-purpose autonomous services in car-dependent peripheries. It works, in other words, because it creates the institutional and spatial capacity to govern whichever scenario materializes first.

The toolkit exists, and it is well proven. Sensing is urban data collection; rules are the commercial regulation of transport operators; economic instruments are curb management, congestion charging, and fiscal planning; design standards are planning policy. Cities already have the tools to do all four. Adapting those tools to the context of CAM will require coordination and the institutional commitment to treat autonomous mobility as a governance priority rather than a technology spectacle.

The competition that matters is not cities against operators, or regulation against innovation. It is between cities that build governance capacity now and cities that wait. The first group will author the operating conditions, revenue structures, and spatial patterns that shape autonomous mobility in their jurisdictions. They will do this on terms consistent with public goals: equitable service distribution, fiscal sustainability, energy coordination, democratic legitimacy. The second group will inherit patterns defined by operators, set in other jurisdictions, optimized for private return. The gap between these outcomes is determined by institutional decisions a city makes before its own thresholds close – the license terms it sets, the consents it grants, and the contract cycles it commits to.

National frameworks settle on known dates. The dates on which a particular city’s options close is unknown, set by that city’s own license terms, consents and contract cycles. But the latter are within the city’s control. A city that has identified its own reversibility thresholds is governing autonomous mobility.

The Urban Autonomy Lab will publish learnings from the first cities to do this work. We invite you to join us and contribute to building the future of urban autonomy.

What To Do Now

8.1The governance window

8.2No-regrets actions: 90 days

Score your Institutional ODD across all four domains

Audit existing authority against the instrument menu

Initiate the minimum viable operating picture

Establish the cross-agency governance wrapper

Engage operators on data-sharing expectations

8.3Building capacity: 12 months

Design the licensing framework

Pilot reversible street and curb interventions

Commission the fiscal transition plan

Coordinate energy and depot planning

8.4Institutional maturity: 3–5 years

Deploy the full operating picture

Implement the economic toolbox at network scale

Embed conversion-ready design standards

Build institutional memory through the recursive learning cycle

8.5What each actor owes

8.6What this paper does not yet contain

8.7The adaptive city

Notes