Ground Truth vol.3 | Building the “Air Traffic Control” for AVs with Autolane
On why the parking lot is the next frontier of autonomous infrastructure, how AVs are flying blind on private property, and the moonshot economics of merging people and packages into a single vehicle.
Welcome back to Ground Truth, a new series from Ride AI that focuses on the entrepreneurs and executives building the infrastructure layer for scaling autonomous vehicles.
This week we’re talking to Ben Seidl, founder and CEO of Autolane, who started the company in 2024 with a conviction that a critical missing ingredient in the growth of autonomous commerce was the orchestration and coordination layer at the curb. The cars, he argues, have gotten remarkably good at thinking and driving. What nobody built was the system to receive them.
Autolane is betting on the seemingly mundane yet fascinatingly complex type of problem we love here at Ground Truth….what happens the moment a self-driving car has to navigate unique road spaces such as crowded parking lots?
Ben, what’s the core idea, and where does that sit in the AV stack?
The idea from the beginning was always centered around one question: how do you enable self-driving cars to actually accomplish things? To do deliveries, interact with a property, handle ride-hailing. Today’s self-driving cars are very good at thinking and driving safely. That’s the extent of what they can do. The obsession became: what if these cars could run your errands? Then all of a sudden, a car in the future would be almost like a personal assistant. Send your car to pick up groceries and drop off a FedEx package. That’s the premise behind Autolane at the highest level.
The first chapter was building the technology that enables self-driving cars to be received on private property in a coordinated fashion. Over the last nine months, that’s rolled out across many of the major shopping centers across the United States — essentially building what is akin to the Uber and Lyft rideshare areas in an airport, but for autonomous vehicles. Working with the property owner to set designated zones. Installing a smart curbside sign at those zones that monitors autonomous vehicle activity, and makes that data available to the property owner so they can start to understand how their property is adapting to this new technology.
Why is the curbside orchestration layer a “need to have” rather than a “nice to have”?
As autonomy scales, there will be a growing need for solutions, tools, and technologies that enable autonomy to interact with society in an efficient way. Because the cars, left to their own devices, are essentially black boxes.
Take the Chase Center in San Francisco. Take a Costco parking lot on a Saturday afternoon. Then imagine what those spaces look like in two years if autonomy continues to scale at the rate it has. You’re probably looking at a very, very chaotic situation. Costco parking lots are already chaotic as they stand today, and that’s with more or less 0% autonomous transportation. What happens when that lot is 10% autonomous and 90% human? As a human driver, you can’t negotiate with one of these vehicles. There’s no one to roll down your window and talk to. There’s no shared language to navigate interactions between AVs and normal vehicles, so it’s going to be an absolute disaster if nothing like Autolane exists.
You describe what Autolane is building as the equivalent of “air traffic control” for parking lots. What does that actually mean in practice?
If you don’t have Autolane today and you’re receiving autonomous vehicles, you are running an airport blind. You can’t communicate with those cars. Those cars won’t go where you want them to go on your own property. You don’t know how many there are, when they’re coming, and which companies are sending them to your property. They’re just showing up unannounced.
That works fine for human drivers who can read signs, navigate, ask questions, and figure things out with context. Self-driving cars can’t. So there are really three layers to solving this properly.
The first is ground truth data monitoring. (Editor’s note— I see what ya did there Ben) Our smart curbside sign doubles as wayfinding for passengers and contains a camera and an AI system that observes and timestamps every vehicle arriving. A store owner can see how many times vehicles are showing up, what types, and for what purpose. That data is the foundation of everything, and it’s the only ground truth a property owner has about what’s actually happening at their curb.
The second layer is rule-setting. Designated stalls, GPS coordinates communicated to the AV companies, so vehicles arrive precisely where they’re needed. If a self-driving car shows up to do a curbside pickup at Home Depot and parks itself on the other end of the lot, the store associate has to walk five minutes across the parking lot to load the order. That’s not scalable. It doesn’t work.
The third layer is protection. If you’ve done everything right and the car shows up to the right stall but a human has parked there—because humans park wherever is most convenient to them, period, at all times—the whole system falls apart. A gate, authorized by the smart curbside sign via license plate recognition, closes that loop. The car arrives, the sign recognizes it, the gate drops, the trunk pops, the order goes in, the car backs out, the gate comes back up. The next reservation runs in three minutes. That’s what a genuinely orchestrated autonomous commerce operation looks like.
Autolane is explicitly OEM-agnostic. In a space where most of the major players are all running proprietary models, how do you operationalize that coordination layer?
In practice, Autolane is a fully approved developer in the Tesla software ecosystem. So we’re building solutions on top of their APIs for trunk control, cabin navigation instructions, speed monitoring, location, and telemetry. On the robotaxi side, the work is giving GPS coordinates to AV companies for the properties that have been activated and saying: update your maps and your pickup and drop-off locations to reflect this going forward. Rather than requiring property owners to have individual conversations with every OEM, Autolane handles all of it on behalf of the property owner. It’s fairly straightforward today, but it will get more complex as self-driving models improve and their reasoning capabilities increase. The integrations will become increasingly programmatic.
Let’s talk partnerships and GTM strategy. Autolane has been working with a wireless charging provider, and there’s clearly a broader ecosystem play here. How do you think about sourcing those mutually beneficial partnerships and, practically, how do you communicate the value to property developers who just want their parking lot to work?
Say a McDonald’s location has Autolane installed. It can receive autonomous vehicles of all types. It can control trunks, coordinate delivery stalls, monitor activity. Autonomous commerce starts flowing. McDonald’s sees the results: lower delivery costs, no commission to third-party marketplaces if they don’t want it, and the ability to do house delivery using their own fleet.
And then the natural next question comes: how do we get more of these cars? That’s where the partnership ecosystem comes in because Autolane isn’t the answer to all four components. Vehicle financing, insurance, fleet management, and charging: those are distinct problems that need distinct partners. What Autolane builds is the network that connects demand to supply. McDonald’s needs 30 cars servicing three stores in Norman, Oklahoma. That order goes into the network. A financing partner underwrites it against McDonald’s delivery history, which is gold-standard data. An insurance partner comes in. A fleet management operator maintains and charges the vehicles.
The underwriting story is what makes this compelling at scale. The largest companies, like Walmart, McDonald’s, Home Depot, and Target, have rock solid delivery volume data. Trends are easy to read, and they may even guarantee contracts. The risk calculus for anyone going out to acquire 30 autonomy-enabled vehicles to service those businesses is very manageable.
For smaller independently owned businesses, like Mary’s Coffee Shop and Don’s Flower Shop, the model shifts. No single small business alone would have the delivery volume to justify a dedicated vehicle. But package three of them together, analyze the combined data as a single entity, and suddenly there’s a viable one-car opportunity for that zip code. Pair businesses whose delivery windows complement each other–the coffee shop heavy in the morning, the flower shop heavy in the afternoon–and that car is busy all day. The shared utilization economics of autonomous commerce are entirely different from what anyone has been able to build with human drivers.
Let’s say I’m a developer trying to future-proof my properties for a future where AVs become ubiquitous. How can I predict what becomes of the humble parking lot and plan my residential or commercial designs accordingly?
The honest answer is: don’t try to predict it. Adapt to it, dynamically. Any significant financial or operational decision made around autonomous vehicles today is almost certainly going to be wrong because no one can predict exactly what will happen in this space. Not even the vehicle companies themselves can reliably forecast their own deployment timelines.
What properties need is not a fixed design decision. They need flexible systems that collect the ground truth data to make the right call at the right time. That’s why Autolane’s hardware is solar-powered, off the grid, has a 4G LTE modem, communicates with the cloud itself, and isn’t even bolted into the ground. It can be moved. The system was designed specifically for adaptability.
Start with one or two stalls. Watch the data. As autonomous vehicle volume grows, you start conversations with your architects and property managers. At what point do you need more stalls? At what point do you break them into different zones? At what point do you dedicate entire areas just to autonomous activity? At what point do you redesign the building around this entirely? Those decisions will come, but they need to be derived from ground truth data rather than speculation. Install early and you start building a picture of how this technology is actually impacting your specific property. And from that, you adapt at pace.
Let’s close with the moonshot version of Autolane in which vehicles simultaneously serve passengers and packages on the same route. How real is that, and what’s the path there?
This is probably the wildest thing that’s been running in the background. Autolane has built custom smart lockers designed to fit fully in the trunk of a Model Y. The back three seats and the front two seats are completely untouched. So imagine this car packed with six orders going out in a sub-two-hour delivery window on a route and simultaneously taking up to five passengers for rideshare.
The car charges the rideshare customer some fare. It charges six different delivery customers a delivery fee. The ride-share interface controls door access. The delivery interface controls trunk access. Neither can reach the other. A passenger never touches the delivery cargo, and a delivery customer never interacts with the people inside. The routing optimizes for both activities.
Why has nobody done this before? Because for a human driver, it’s insane. There’s too much to manage, too many variables, no way to optimize. With autonomous vehicles, it’s all code. And the economics of it are extraordinary. The marginal cost of adding a ride-share passenger to a vehicle already running a delivery route could push the fare close to zero. A dollar for ten miles. Something that functions like a future bus with dynamic routing executed for nearly nothing. That’s the moonshot, and it’s closer than we think. The CyberCab, for example, is already at the equivalent of 204 miles per gallon, and these wildly electrically efficient vehicles will operate inside an orchestration network that pushes utilisation toward 90 or 95%. When that happens, the cost of moving both people and goods drops to something previously unimaginable. And the number of vehicles required to serve all of that demand falls dramatically. That’s what the curb is capable of enabling.
Ground Truth is a series from Ride AI focused on the entrepreneurs and executives building the infrastructure layer for scaling autonomous vehicles. To suggest candidates or discuss partnership opportunities, reach out to mike@rideai.org.