The Robotaxi Should Be Waiting When the Train Arrives

Imagine stepping off a train after a long day. You have a bag over your shoulder, your phone in your hand, and one more connection before you get home.

Now imagine two versions of what happens next.

In the first, you request a ride and watch a vehicle work its way toward the station. You follow a map to a pickup point, check whether you are standing in the right place, and wait.

In the second, vehicles have been positioned nearby ahead of the train’s arrival. You request your ride, follow clear directions to an appropriate meeting point, and meet your vehicle.

That second experience is what we should be working toward.

Robots should wait for people. Not the other way around.

That is not a promise that nobody will ever wait for a ride. It is a principle for deciding how to organize the service: start with the person’s journey, then work backward to where the vehicle needs to be.

Start with the passenger, not the parking space

When evaluating a location for autonomous vehicles, it is easy to start with the property: How many spaces are available? Can vehicles enter? What does it cost?

Those are necessary questions. But they are not the first questions.

The first question should be: Does this location make the passenger’s trip easier?

A space can be convenient for a fleet and inconvenient for the person trying to reach it. A nearby lot might require an awkward crossing, an unclear entrance, or a walk that is unreasonable for someone carrying luggage, accompanying children, or using a mobility aid.

One robotaxi operator’s own pickup guidance recognizes that road closures, traffic, obstructions, and stopping restrictions can prevent a vehicle from reaching a passenger’s exact location. It also notes that minimizing walking can involve a tradeoff with ride time. The pickup experience is not simply a matter of sending a car to a pin.¹

Consider a hypothetical pickup: the app says the vehicle is two minutes away, but the passenger needs eight minutes to reach the meeting point.

That is not a two-minute pickup.

A short walk may be worthwhile when it produces an easier, more reliable departure. But we should evaluate that tradeoff from the passenger’s perspective, not congratulate ourselves because the vehicle’s arrival estimate improved.

Give vehicles somewhere useful to wait

Staging means positioning vehicles where they can wait before they are needed.

On September 22, 2026, a robotaxi company announced plans to lease 40 dedicated parking spaces at Caltrain stations to stage vehicles for transit riders. That is an announced arrangement, not evidence by itself of shorter passenger waits, but it provides a concrete example of planning vehicle capacity around a transit connection.²

The opportunity is to plan around expected arrivals, update that plan as conditions change, and have appropriate capacity nearby.

That does not mean leaving unlimited vehicles idle at every station. It means treating waiting capacity as part of delivering the service rather than an afterthought.

There is another important distinction: where a vehicle waits does not have to be where a passenger boards.

A vehicle could wait in an approved nearby space and move to a designated pickup point when the rider is ready. Passengers should not have to navigate a storage area or service yard simply because that was the easiest place to accommodate cars.

The staging location should support the pickup experience, not dictate it.

Plan the departure before the event ends

The same principle applies to stadiums, concerts, and other major events.

Imagine leaving a concert with friends. You do not want to study a traffic-management map. You want to understand where to go, find your ride, and get home.

A useful plan should account for that experience before the crowd leaves.

Where can vehicles wait in advance? How will passengers find the meeting point? Is the walking route reasonable and accessible? Can vehicles reach the pickup area under the event’s traffic arrangements? Once passengers board, can the vehicle leave without being directed back into the congestion?

The nearest available lot is not automatically the best answer.

A somewhat more distant pickup may offer a better departure, but only when the walking route and the complete journey justify it. Moving inconvenience from the vehicle to the passenger is not a solution.

There also needs to be an alternative when the original plan fails. A blocked entrance or changed traffic pattern should not leave passengers trying to invent their own meeting point.

Measure the whole human journey

If we say this is human-first, our measurements should reflect it.

We should measure the time from leaving the station or venue to boarding, not only the time after a vehicle is assigned. We should understand how much walking was required, whether directions were clear, whether the pickup succeeded, and whether passengers could comfortably use the route and boarding point.

Then we should look at what happened after boarding. Did the vehicle depart efficiently, or did a seemingly convenient pickup lead straight into a difficult exit?

Averages are not enough, either. A journey that works well for many passengers may still be impractical for someone with different accessibility needs.

The test is not whether the car arrived faster. It is whether the person had a better journey.

Give local properties a useful role

This is also an opportunity to involve the places and businesses already around stations and venues.

A participating property could provide suitable waiting capacity during agreed hours. The owner should understand the permitted use, retain control over the arrangement, and have a commercial reason to participate.

But an available space is only the beginning. Access, capacity, operating hours, passenger routes, pickup instructions, and backup arrangements all need to work together.

That is the principle we are building around at XoomPark through XoomPark OS: connect suitable properties and clear operating arrangements to the needs of fleets and the people they serve.

The software is a means to that end. The outcome should be easier to recognize: a passenger knows where to go, a vehicle is positioned to meet them, and the participating property knows what to expect.

For a station, venue, or nearby property owner, the starting question should not be, “How many robotaxis can we fit here?”

It should be:

“What would make it easier for someone to get home, and what needs to be in place before they arrive?”

That is where human-first autonomous mobility begins.