EMBRACE THE PARADOX: Urban traffic and robotaxi orchestration

Bern Grush, Executive Director of the Urban Robotics Foundation, talks to Intertraffic about the coming revolution – driverless taxis on our cities’ streets… or more specifically, how they will operate in and with our cities.

On a February bike ride in 2025, 26-year-old San Franciscan Jenifer Hanki was in a bike lane when a robotaxi pulled into her path to drop off passengers. As she attempted to pass, a passenger opened its rear door, throwing her into a second robotaxi, also in the bike lane. The impact left Hanki with brain injuries and spinal damage. This crash encapsulates a growing crisis that threatens the autonomous transport sector: the fundamental challenge of orchestrating pickup and drop-off operations.

For Bern Grush, Executive Director of the Urban Robotics Foundation based in Toronto, the unfortunate incident involving Jenifer Hanki highlights just how much more work still has to be done before driverless vehicles are as commonplace a sight in our cities as human-controlled cars.

“It is clear that robotic taxis are performing remarkably in several cities,” he says. “While the realisation grows that they will never be perfect, it is also clear that many of us will use them. For those who do, the highlight of our trips will fade, as we will be busy on our smartphones or even napping. Instead, the focus of our trips will be about getting picked up and dropped off.”
 

“For those who use robotaxis the highlight of our trips will fade, as we will be busy on our smartphones or even napping. Instead, the focus of our trips will be about getting picked up and dropped off.”

Waymo by Elliott Coward Jr. - Dreamtime

If you use ride-hailing now, you understand the variability of positioning yourself to be picked up and hoping you will be dropped off conveniently, safely, and close to your destination. Sometimes, however, you will get in or out while the taxi stands in a bicycle lane or in a specific traffic lane, possibly blocking a bus, ambulance, or firetruck. Getting from A to B will be focused far less on the trip, and almost solely on its endpoints. Congestion, safety and comfort at the trip's endpoints will rival the importance of your safety during the trip itself.

“The robotaxi industry's Achilles heel is not its sophisticated AI or sensors, but in its passengers’ experience when stopping for them to get in or out,” Grush maintains. “Unlike human drivers who make dynamic, opportunistic decisions about where to stop, automated vehicles require reliable, reserved spaces. Without proper coordination, pickup and drop-off threaten worse chaos than cities have experienced in 125 years of kerb management.”
 

The robotaxi industry's Achilles heel is not its sophisticated AI or sensors, but in its passengers’ experience when stopping for them to get in or out.

Growing scale

Industry projections suggest major urban areas will see thousands of automated vehicles by 2030, each generating approximately 50 daily pick-up and drop-off (PUDO) events. A modest fleet of 1,200 robotaxis implies 22 million PUDO events annually in a single city. Current municipal kerb management systems, already strained by ride-hailing and e-commerce deliveries, are just not ready to handle such an increase.

“Removing human drivers fundamentally changes kerb access dynamics, even as demand grows. Human drivers make eye contact, negotiate through gestures, or choose alternative stopping locations,” Grush explains. “Automated vehicles require explicit instructions and algorithms. This shift from human improvisation to algorithmic precision creates new urban mobility challenges.

‘Unmanaged PUDO operations for robotic vehicles pose significant safety risks. Automated vehicles without assigned PUDO spots resort to stopping in traffic lanes, bike lanes, or other unsafe locations. The San Francisco incident demonstrates how operational gaps translate into real harm for vulnerable road users. The robotaxis’ opportunistic decisions to use the bike lane illustrate how current kerb management approaches are inadequate for automated vehicle operations.”
 

Human drivers make eye contact, negotiate through gestures, or choose alternative stopping locations. Automated vehicles require explicit instructions and algorithms.

Continuous operation

Economic models driving robotaxi deployment exacerbate these challenges. Without driver labour costs, robotaxis can operate almost continuously, generating more daily PUDO events than conventional vehicles could. This creates a paradox: the efficiency making robotaxis commercially viable threatens to overwhelm the infrastructure they depend upon, as Grush elucidates.

“Traditional taxis complete 20-30 trips daily with significant downtime. Robotaxis can operate 20 hours daily with minimal maintenance breaks, multiplying pressure on contested kerb space. Competition further complicates coordination. Uber doesn't coordinate its trip plans with other ride-hail operators, nor does FedEx with UPS. Similarly, automated fleet operators will independently optimise vehicle operations without regard for system-wide efficiency unless required to participate in an orchestration system.”

Accessibility

PUDO challenges intensify when accessibility requirements are considered. Wheelchair users require designated stopping space with additional room for ramp deployment and safe manoeuvring. Visually impaired passengers, on the other hand, need consistent, predictable pickup locations.

“Current kerb management systems treat accessibility as an afterthought, with designated spaces that may be unavailable when needed. For robotaxi operations, this becomes untenable. Automated vehicles cannot improvise alternative arrangements when accessibility spaces are occupied,” he insists. “Accessibility must be built into the fundamental architecture of automated vehicle kerb management systems.”
 

Current kerb management systems treat accessibility as an afterthought, with designated spaces that may be unavailable when needed. For robotaxi operations, this becomes untenable.

 

Ground traffic control

The PUDO challenge resembles managing aircraft landings. When multiple aircraft approach airports simultaneously, they rely on central coordination rather than negotiating among themselves. Similarly, PUDO orchestration systems coordinate spaces for ground vehicles that require kerb access. All participating fleets communicate with central systems that allocate kerb space based on real-time availability and vehicle needs.

“Draft international standard ISO 25614 is in preparation to address this,” adds Grush, one of the standard’s driving forces. “It defines message formats and data flows to coordinate curb access without requiring vehicle operators to expose proprietary information. Transitioning to robotaxi operations demands fundamental reimagining of how cities manage kerb space. A 2018 International Transport Forum study found that in European and North American cities, significant portions of kerb space were allocated to private vehicle parking, despite this representing one of the least efficient uses of valuable resources.”

This inefficient allocation becomes unsustainable in a robotaxi future.

“Cities must reapportion static parking allocations to dynamic space allotments in response to changes in demand. Such management must account for varying operational requirements: some package deliveries take longer than passenger drop-offs; passengers with mobility challenges need longer boarding times.”
 

Cities must reapportion static parking allocations to dynamic space allotments in response to changes in demand.

 

Waymo by Elliott Coward Jr.

Inaction

The vicious cycle impacts of inadequate kerb management are well documented. Without adequate loading zone supply, drivers spend more time searching for parking or resort to using unauthorised spaces. These behaviors reduce roadway capacity and add spatial conflicts.

Cities failing to address PUDO challenges risk being overwhelmed by the same automated technologies meant to improve transportation, says Grush.

“Without orchestration, even modest-sized automated vehicle fleets compete for kerb space, double park, increase congestion, add safety hazards and compound enforcement issues. These problems, already evident with current vehicle volumes, will intensify exponentially as robotaxi fleets scale. Without coordinated PUDO management, cities risk having transportation systems that are more chaotic than today's.”

Embracing the paradox

The path forward requires embracing a second key robotaxi paradox: individually, robotaxis behave better than human-driven cars but collectively they require external coordination. PUDO management provides centralised orchestration where all automated vehicles participate to prevent chaos and optimise space utilisation.

“Orchestration systems must handle multiple complex requirements simultaneously: real-time space availability tracking, advance reservations for specific time slots, dynamic reassignment during delays, special accessibility accommodations, priority management for different vehicle types, and coordination with traffic flow patterns…” Grush pauses, somewhat impressed with his improvised checklist.

“The San Francisco crash that severely injured Jenifer Hanki reminds us that technological advancement without proper infrastructure coordination creates new forms of harm. The robotaxi industry's Achilles heel - managing pickup and drop-off operations - represents both the greatest challenge and opportunity in urban transportation planning.”
 

The San Francisco crash that severely injured Jenifer Hanki reminds us that technological advancement without proper infrastructure coordination creates new forms of harm.

Cities engaging PUDO orchestration systems will realise automated mobility's full benefits: congestion minimisation, improved accessibility, enhanced safety, urban space efficiency and, crucially, monetisation recovery.

“The choice facing city planners and traffic engineers isn't whether automated vehicle deployment will be significant - it will,” states Grush with confidence. “The choice is whether to optimise kerb management or react to unmanaged automation consequences. Just as we wouldn't deploy thousands of aircraft without air traffic control, we simply cannot realise automated ground transport's promise without addressing fundamental kerb use coordination.
 

“The robotaxi industry's Achilles heel, as I mentioned earlier, need not be its downfall,” he concludes. “With proactive planning, comprehensive coordination systems, and collaborative governance models, this challenge can become the foundation for more efficient, equitable and sustainable urban transport systems. The question isn't whether cities can afford to invest in PUDO orchestration - it's whether they can afford not to.”