Robotaxis: Chaos or coordination ?

How will cities manage the burgeoning proliferation of robotaxis? What are the priorities? Who gets the kerb? How is the 'mobility revolution' actually going to work? Kevin Borras quizzes Bern Grush

The announcement that Waymo is to begin its driverless taxi service in London, UK, was met with, it has to be said, a mixed response. Excitement and intrigue in some quarters, while in others there was concern for not only the livelihoods of the capital's taxi drivers, but also consternation about how the autonomous vehicles (AVs) will behave on the city's notoriously congested road network.

The answer to the behaviour question is already being addressed by a Canadian start-up, co-founded by a man whose name (and thinking) is very familiar to the ITS community.

"The robotaxi revolution is transforming urban mobility, but without proper coordination driverless vehicles are creating new problems," says Bern Grush, at least half of the brains behind Pudocity, purveyors of what is, in other words, 'orchestration as a service'.

To address this, ISO draft standard 25614, which has been in process for a few years, appears to be nearing completion. Ontario-based Pudocity has established a team and an architecture to deploy a system based on this communication and data standard. The mission: to create an orchestration platform to solve the challenge of where AVs pick up and drop off passengers or goods. Pick-up/drop-off: PUDO.

The premise: unlike human drivers who are able to make dynamic, opportunistic decisions about where to stop, automated vehicles require consistent, reliable, reserved spaces. Without proper coordination, pick-up and drop-off threaten worse chaos than cities have experienced in 125 years of kerb management. It's a future problem that needs solving now.

"The safety implications of uncoordinated robotaxi operations are already visible in cities where AVs operate," says Grush.

Multiple AV parking violations, for example - including citations for blocking street cleaning zones - aren't just nuisance infractions; they represent genuine safety hazards.

"Something interesting I have just learned is that in as many cities as not, the effective authority will be the parking authority because that is who is currently monetising the kerb," says Grush.

"Since PUDO requires kerb space, this will decrease the available space for parking, and therefore parking revenue. Because our system provides for monetisation, those cities that charge for parking - which is most cities - it's their parking authority that will often be the appropriate decision maker."

Michael Brooks, executive director of the Center for Auto Safety, has emphasised that when robotaxis obstruct traffic flow, they force other drivers to brake suddenly or swerve unexpectedly, increasing crash risk throughout surrounding areas.

Robotaxis have also stopped and blocked roadways, and there have been incidents where AVs impeded police and fire response, including running through emergency tape and blocking firehouse driveways.

Grush explains: "Without human judgment to navigate ambiguous situations, these vehicles need explicit PUDO coordination to avoid creating congestion problems that can cascade across an urban transportation network."

Effective orchestration requires integrating multiple data sources within a comprehensive management system.

The platform has to maintain digital inventories of reservable spots with multi-dimensional spatial and feature specifications, operational parameters such as maximum vehicle dimensions and weight limits, and temporal availability schedules.

Dynamic PUDO spot availability changes constantly based on construction activities, maintenance requirements, emergency situations, special events and temporary restrictions.

"Real-time traffic condition data helps predict optimal arrival times and buffer periods between sequential reservations," Grush points out.

"Historical usage patterns reveal demand fluctuations across different times of day, weather conditions, and seasonal variations. Machine learning models trained on accumulated data can predict dwell times, identify potential conflicts before they occur, and continuously refine assignment algorithms. Priority information ensures emergency vehicles, public transit, and utility services receive preferential access when needed. By synthesising these diverse data streams, the orchestration system can make intelligent decisions that optimise system-wide efficiency rather than simply matching vehicles to the nearest available spot."

Pudocity's orchestration system promises to move beyond conventional proximity-based parking solutions that have proven inadequate at scale. Traditional closest-match approaches fail to account for spatial conflicts when multiple vehicles converge on adjacent spots, temporal issues when sequential reservations are scheduled too closely together, or the broader system impacts of concentrated demand in high-traffic areas.

The Pudocity platform evaluates potential assignments based on current spatial density, temporal conflicts with existing reservations, historical usage patterns, and the impact on system-wide efficiency, Grush explains: "The system can designate a spot slightly farther from the requested location if doing so reduces bottlenecks, maintains uniform density, or keeps high-demand spots available for priority vehicles. This intelligent allocation prevents the natural clustering that inevitably occurs when each vehicle independently seeks the closest available space."

Two critical dimensions of Pudocity's orchestration promise involve managing conflicts in both space and time. When robotaxis are scheduled to arrive at adjacent spots near-simultaneously, the vehicle manoeuvring required to settle can slow or block traffic and create congestion that ripples through the local area.

The system identifies these potential conflicts and proactively assigns vehicles in ways that minimise these effects to dramatically improve neighbourhood traffic flow.

For temporal management, the platform uses predictive models to determine optimal buffer periods between sequential reservations at the same spot, accounting for vehicle size, cargo type, passenger needs, time of day, and even seasonal factors like snow accumulation that can slow parking manoeuvres.

This prevents situations where delayed departures and early arrivals force incoming vehicles to wait or double-park in travel lanes, blocking through traffic.

"The point here is that not all vehicles have equal claims to limited kerb space, and effective orchestration must reflect such legitimate hierarchies," adds Grush, who is also the founder of the Urban Robotics Foundation and, in the early 2000s, was the public face of satellite tolling solutions provider Skymeter.

"The Pudocity system implements priority schemes that ensure emergency vehicles receive immediate access during crises, public transit buses maintain reliable access to designated stops, and construction or utility vehicles can park near work sites," he continues.

"When an ambulance or a fire truck needs immediate kerb access, the system identifies lower-priority reservations nearby and reassigns them to alternative locations - enabling emergency vehicles to arrive at required locations even before they reach the scene."

For public transport, the system maintains buffer zones around bus stops, preventing robotaxis from encroaching on transit infrastructure during scheduled arrival windows.

This coordinated approach addresses one of the most significant current problems: AVs do not reliably respond to gestured instructions from emergency personnel or transit operators.

Grush insists that economic implications extend far beyond avoided parking tickets. For robotaxi operators, eliminating time spent circling for available spots translates directly to increased vehicle utilisation and revenue - a robotaxi that spends five minutes less per trip searching for locations can complete significantly more trips per day.

"Cities stand to benefit even more dramatically through increased utilisation of existing kerb space, dynamic pricing mechanisms that discourage inefficient behaviours, and the recovery of monetary value from historically under-priced public space," he maintains.

"Environmental benefits accompany these economic gains. We know that reduced circling means lower emissions, less congestion means improved air quality, and more efficient operations mean better energy utilisation across the entire fleet. Pudocity's orchestration system promises to capture value that currently goes unrealised while simultaneously improving transportation efficiency and environmental outcomes."

Perhaps the most powerful promise of the Pudocity platform is its ability to learn and improve over time. Machine learning models trained on historical usage data identify patterns invisible to human planners: which locations experience highest demand during different times, how weather affects dwell times, what buffer periods are optimal for different vehicle types.

The system detects when particular spots consistently generate delays and adjusts assignment decisions accordingly, identifies underutilised locations and load balances within local regions - and even learns inter-fleet operator differences to improve reliability.

As AV deployments scale from dozens to thousands of vehicles across metropolitan regions with multiple competing operators, regional orchestration becomes not merely beneficial but essential.

Grush simplifies it like this: "No individual operator can solve system-wide coordination problems, and only a neutral third-party manager with visibility across participating fleets can optimise for collective benefit rather than individual advantage."

The AV industry stands at a critical juncture, where technical capability has already outpaced operational coordination.

"Pudocity's orchestration system, built on the emerging ISO 25614 standard, represents the missing infrastructure layer to unlock the full potential of autonomous mobility," states Grush.

"Just as traffic signals coordinate vehicle movements and air traffic control manages aircraft landings on shared runways, this platform promises to coordinate access to limited urban infrastructure before uncoordinated growth creates unmanageable problems. The technology exists, the economic case is compelling, and the safety benefits are substantial. The question isn't whether robotaxis will transform our cities, but whether that transformation will be chaotic or coordinated."