Autonomy Doesn't Stop at the Curb
A driverless ride economics deck usually focuses on driving cost per mile. The chart that gets less attention is what happens between rides — the cabin reset that, in a human-driven service, the driver does silently while waiting at a light.
In a driverless vehicle, that reset has no one to do it. Today the answer is depot returns and human cleaners. That works at small scale. It is a structural cost as fleets grow.
What Riders Actually Complain About
Public rider feedback for the major driverless services consistently calls out the same handful of issues: spilled drinks, food residue, lingering smells, and occasional belongings left behind. None of these are autonomy bugs. All of them shape repeat-ride behavior more than the driving itself does, because the driving has become unremarkable in a good way.
Where to Put the Cleaning Step
Two operational patterns are emerging. Depot-based cleaning does a deep reset at scheduled returns and is easier to staff (or, in our case, automate). On-route micro-stops would route a vehicle to a quick-clean station between problematic rides; this requires tighter dispatch integration and a station footprint at the metro edges.
Both have a place. Depot handles the scheduled deep reset; on-route handles the spilled-coffee exception. An autonomous cabin platform that can do both at depot scale is the wedge.
Talking to Dispatch
For this to work the cleaning system has to be a first-class citizen in the dispatch stack: a vehicle's 'available' state must depend on its 'cleaned' state, and exceptions need to flow back so dispatch knows to hold the car. This is unglamorous integration work; it is also what separates a pilot from a deployment.
Book a robotaxi pilot conversation
Visit handybot.ai →Leather, alcantara, vinyl, fabric, glass, rubber mats — and why a SAM2 baseline gets fooled by reflections.