Regulatory Landscape: Which Cities Allow Sidewalk Delivery Robots in 2026
State authorization and city authorization are not the same thing — and the gap between them is where operator risk lives.
When Starship Technologies began campus operations in Pittsburgh, the University of Pittsburgh paused the program after a single incident: a wheelchair-using student posted that a robot had partially blocked a curb ramp. The program resumed only after Starship updated the robot's software to improve its behavior around accessibility infrastructure.
That episode captures the regulatory situation for sidewalk delivery robots in 2026 better than any piece of legislation: state law permitted the operation, university policy permitted the operation, the technology mostly worked — and a single accessibility incident was sufficient to halt everything pending a policy review.
If you're planning a sidewalk delivery deployment, you need to understand the regulatory environment at three levels: state enabling law, municipal permitting, and ADA compliance liability. Each level operates independently, and clearance at one level does not provide clearance at the others.
Level 1: State Enabling Law
As of 2026, five U.S. states have passed legislation that explicitly permits autonomous delivery robots to operate on public sidewalks: Pennsylvania, Virginia, Idaho, Florida, and Wisconsin. Several others are in various stages of consideration.
What state enabling law typically does:
- Classifies delivery robots as "pedestrians" for traffic law purposes, giving them right-of-way on sidewalks
- Sets speed limits (typically 10–12 mph maximum on sidewalks, lower in pedestrian-dense areas)
- Establishes weight limits (most statutes cap at 50–80 lbs)
- Requires that operators maintain liability insurance
- May require that a human be capable of taking remote control
What state enabling law typically does NOT do:
- Override municipal zoning or permit requirements
- Pre-empt city ordinances on sidewalk use
- Establish ADA compliance standards (those come from federal law)
- Define who is liable in the event of a pedestrian incident
In states without enabling legislation, the legal status of sidewalk robots is ambiguous — they may be classified as vehicles (prohibited from sidewalks), as pedestrians (permitted), or as something undefined (subject to case-by-case regulatory interpretation). Operating in a non-enabling state is a legal risk that vendors will often present as manageable; their assessment of "manageable" may not match yours.
Level 2: Municipal Permitting
Even in states with enabling legislation, cities retain meaningful authority over sidewalk operations. The regulatory friction that states like Pennsylvania broadly permit can coexist with the local friction that cities like Pittsburgh actually impose.
San Francisco
Requires permits for each deployment zone. The city's process is among the most structured in the U.S. — permits specify zones, hours of operation, speed limits within those zones, and insurance minimums. Serve Robotics has operated in San Francisco under this framework. The permit process takes weeks to months, and zone approvals can be contested by neighborhood groups. San Francisco is permissible but not easy.
Los Angeles
Serve and Coco operate in Los Angeles. The city has not established a citywide permitting framework as formalized as San Francisco's; deployment has proceeded under a combination of state law and case-by-case negotiation. This is a common pattern in large California cities: the enabling environment exists, but the rules are being made up in real time.
Miami
Active deployments: Serve, Coco, Cartken (via Uber Eats). Florida's enabling legislation is among the more permissive in the country, and Miami has been a growth market for multiple operators simultaneously. The warm climate, dense restaurant corridor around Brickell and South Beach, and state-level support have made Miami one of the more accessible U.S. markets for city deployment.
Dallas-Fort Worth
Serve Robotics launched in DFW in April 2025, covering 22,000 households via Uber Eats. Texas has been relatively permissive at the state level; DFW's suburban geography creates different challenges (lower pedestrian density, greater distances between stops) than a dense urban grid, but fewer regulatory obstacles.
Houston
Active for Nuro/Uber Eats food delivery partnerships. Texas enabling environment applies. Houston's sprawl means robot deployment is concentrated in specific neighborhoods rather than city-wide.
Pittsburgh
Technically permissive under Pennsylvania law. The university incident referenced above illustrates how community pressure can operate independently of legal authorization. Pittsburgh's disability advocacy community is active and engaged — any deployment in the city needs a robust ADA plan that goes beyond the legal minimum.
Chicago
Coco operates in Chicago. The city has seen more community friction around delivery robot operations than most U.S. markets — concerns about sidewalk privatization, pedestrian displacement, and gig worker displacement have generated organized opposition in several neighborhoods. Chicago is permissible but politically contested.
New York City
As of 2026, no major sidewalk delivery robot operator has achieved sustained city-scale deployment in New York. The density of pedestrian traffic, the contested nature of sidewalk real estate, and the political environment around gig economy labor regulation make New York one of the most difficult markets in the country. Several operators have tested here and pulled back. Do not plan a New York deployment without a city-specific regulatory strategy.
Level 3: ADA Compliance
The Americans with Disabilities Act is federal law, and it applies regardless of state or municipal enabling legislation. Sidewalk robots that operate on public sidewalks must not impede the mobility of people with disabilities.
In practice, ADA compliance for sidewalk robots means:
- Yielding to wheelchair users and mobility aids
- Not blocking curb ramps (the Starship Pittsburgh incident)
- Not operating in ways that create hazards for visually impaired pedestrians
- Maintaining accessible path widths around the robot
Most vendors will tell you their obstacle avoidance algorithms handle this. The honest answer is that they handle it most of the time, in environments the algorithm has been trained on. The failure modes — a robot that hesitates at an unexpected obstacle configuration, a robot that stops in a way that partially blocks a ramp — are low-frequency but high-consequence.
The liability picture is not fully defined. A plaintiff's attorney could argue that a business deploying a sidewalk robot is responsible for any ADA violation the robot commits, regardless of whether it was a vendor system failure or an operator configuration issue. In a market where the case law is still developing, that ambiguity runs against the operator.
The minimum defensible practice: before deploying in any corridor, walk it with a disability access consultant and identify every curb ramp, tactile warning surface, and accessible path of travel. Provide that map to the vendor and get written confirmation that the robot's behavior in those zones has been specifically tested. Keep that documentation.
How to Assess Jurisdictional Risk
Before committing to a deployment location, run through this checklist:
State level:
- Is the state an enabling state? (PA, VA, ID, FL, WI, and others in progress)
- What does the enabling statute specifically permit and restrict?
- Are there weight, speed, or operational hour restrictions in the statute?
Municipal level:
- Has the city established a formal permitting process?
- What are the permit conditions (zones, hours, insurance minimums)?
- Has the vendor previously deployed in this city? What was their experience?
- Are there active neighborhood or community organizations that have contested robot deployments?
- What is the city council's current posture toward sidewalk delivery?
ADA level:
- Has the deployment corridor been assessed for accessibility infrastructure?
- Has the vendor provided documentation of ADA-specific testing on similar infrastructure?
- Who carries ADA liability — vendor or operator?
- What is the vendor's incident response process for accessibility complaints?
Risk summary: Cities that combine enabling state law + a formal permitting process + low pedestrian density + no active community opposition are the lowest-risk deployments. Miami, Dallas-Fort Worth, and campus environments fit this profile. Chicago, San Francisco, and Pittsburgh carry higher operational risk requiring more robust community and compliance preparation. New York remains a frontier market.
What's Coming in 2026–2027
Regulatory activity is accelerating. Cities that have not yet established frameworks are being pushed to do so as operators seek to expand beyond their initial geographies. The general direction is toward more formalized permitting (following San Francisco's model) rather than blanket prohibition — but the form that permitting takes varies enormously by city.
The most significant pending question is federal: whether NHTSA, the FTC, or DOT will establish a national baseline for sidewalk robot safety and liability that pre-empts the current patchwork. Several industry groups have submitted comments supporting a federal standard; there has been no binding rulemaking as of 2026.
For operators planning 18-to-24-month deployment horizons, the practical strategy is to launch in permissive geographies, build a compliance track record, and use that record in permit applications in more regulated markets. A vendor with zero incident reports across 50,000 deliveries in Miami has a stronger argument before the San Francisco permit board than a vendor with no field history.
The next article in this series covers the cost gap between indoor and outdoor delivery — because the regulatory risk for hospital and hotel indoor deployments is a completely different picture.
