Stanley Tang on DoorDash DOT: 7 years to build, L4 autonomous delivery live in Phoenix today

Summary

DoorDash co-founder Stanley Tang used this week's announcements to unveil DOT, the company's autonomous delivery robot, after seven to eight years of development inside DoorDash Labs.

DOT is designed specifically for the 3-to-5 mile suburban delivery corridor — the market DoorDash built its business on. At roughly 350 pounds and capable of up to 20 mph, it sits between a sidewalk robot and a full-size autonomous vehicle: a tenth the footprint of a car, but fast enough to be commercially useful. It travels on bike lanes, roads, and sidewalks. As of the announcement, DOT is live in Phoenix at full L4 autonomy, handling real deliveries without human oversight.

The delivery-specific problem

Tang argues that autonomous delivery is a fundamentally different engineering problem than autonomous ride-share. A robo-taxi moves people between two points. DOT has to navigate merchant pickup, the first and last 50 feet of both ends of a trip, and the operational logic of wildly different order types — restaurant food, groceries, ice cream, iPads. DoorDash's 10 billion lifetime deliveries give it precise data on pickup and drop-off points, building-entry logic, and distance profiles that no robo-taxi company has needed to solve.

Multimodal fleet vision

DOT is one piece of a broader autonomous delivery platform DoorDash also announced. The intended architecture is modality-matched to use case: human Dashers for complex grocery pick-and-pack or apartment navigation, drones for fast rural deliveries, sidewalk robots for dense urban cores, and DOT for dense suburbs. Tang frames this as a routing and fleet-management challenge — exactly where DoorDash's existing operations capability gives it an edge over pure-play hardware companies.

Technology stack

DOT runs cameras, lidar, and radar, with a camera-primary approach that Tang expects to become more dominant as sensors and ML improve. The broader shift he points to is from heuristics and rule-based autonomy a decade ago to end-to-end ML, a transition he credits Tesla with leading. Waymo's commercial viability in autonomous ride-share was the proof point that changed the framing from R&D problem to commercialization problem — which is where DoorDash believes its operational DNA becomes the differentiator.

Humanoid robots

Tang is open to humanoids in the delivery chain but insists the question has to start with the use case, not the hardware. He sees potential fits at DashMart micro-fulfillment centers or the last 100 feet inside a building. A four-mile outdoor delivery run is not the right application.