When evaluating autonomous sidewalk delivery systems, the milestone announced by Starship Technologies in early 2025 offers a useful reference point for what is actually achievable in this segment today. The company reported that its fleet had completed more than 8 million autonomous deliveries and traversed over 10 million miles globally. These figures, while impressive on their own, become more meaningful when you break down what they imply about operational maturity, system reliability, and the practical realities of deploying sidewalk robots at scale.
The first thing to examine is the fleet size and geographic footprint. Starship operates over 2,000 battery-powered robots across the United States and Europe. That is not a small pilot program; it is a commercially active network serving university campuses and urban neighborhoods. When you are looking at a provider, ask whether they can point to a comparable fleet size and whether those robots are actually in daily service rather than sitting in warehouses awaiting demonstrations. A fleet number without operational context tells you little. The fact that Starship’s robots are described as a familiar sight for thousands of customers suggests sustained, visible operations rather than occasional showcases.
Another critical indicator is the volume of road and driveway crossings. According to the company’s announcement, its robots cross 125,000 roads and driveways per day, which works out to roughly two crossings per second. This is a revealing metric because it speaks to the complexity of the environments these machines navigate. Sidewalk delivery is not just about rolling along a flat path; it involves negotiating intersections, driveways, curbs, pedestrians, and other obstacles. If a system can handle that frequency of crossings safely and reliably, it demonstrates a level of environmental perception and decision-making that is worth scrutinizing when comparing vendors.
The delivery mix is also worth noting. Starship’s robots handle groceries, takeout, and parcel delivery. That diversity matters because it indicates the platform is not narrowly optimized for one use case. A robot that can switch between a hot meal, a bag of groceries, and a small package is more versatile for operators who want to maximize utilization across different demand patterns. When you are assessing a potential deployment, consider whether the provider has proven experience across multiple delivery categories or only in a single, controlled scenario.
Environmental impact is another factor that increasingly influences procurement decisions. Starship states that its electric robots have saved hundreds of tonnes of CO₂ by reducing car usage. While the exact methodology behind that calculation is not disclosed in the source material, the claim aligns with the fundamental logic of replacing short car trips with electric sidewalk robots. If sustainability targets are part of your organization’s mandate, you should ask any provider for their own emissions data and the assumptions behind their calculations. The absence of a standardized metric across the industry means you will need to compare methodologies as much as numbers.
Finally, look at the technological architecture. The source material notes that most delivery bots and robotaxis today are level-4 autonomous vehicles. That means they can drive themselves under specific conditions, such as mapped areas and good weather, but may request remote human help in rare cases. This is a crucial distinction. Level-4 does not mean fully driverless in every conceivable situation; it means the system has a defined operational design domain. When you are planning a deployment, you need to understand what those conditions are for the specific provider you are considering. Ask about their sensor suite—cameras, lidar, radar, and AI software are commonly used—and how they handle edge cases. The source material also mentions that advances in electric batteries allow these bots to operate for tens of kilometers per charge, and that upgrades in 5G and V2X networks improve remote connectivity and data collection. These are practical considerations that affect range, uptime, and the ability to monitor fleets in real time.
Practical steps
If you are considering deploying autonomous sidewalk delivery robots, the Starship milestone provides a template for what a mature operation looks like. Here are practical steps to guide your evaluation and implementation process.
Start by defining your delivery use case precisely. Are you looking to serve a university campus, an urban neighborhood, or a mixed-use district? The source material indicates that Starship’s robots serve university campuses and urban neighborhoods for grocery, takeout, and parcel delivery. These environments have different traffic patterns, pedestrian densities, and regulatory landscapes. A campus may have controlled access points and predictable schedules, while an urban neighborhood introduces more variables such as public roads, driveways, and unpredictable foot traffic. Your use case will determine the specifications you need, from battery range to sensor capabilities.
Next, assess the provider’s track record using the metrics that matter. The 8 million delivery milestone and 10 million miles traversed are headline numbers, but the more granular data—125,000 road and driveway crossings per day, two crossings per second—tells you about operational intensity. Ask any provider for comparable figures. If they cannot provide daily crossing counts or per-day delivery volumes, that is a red flag. You want a system that has been stress-tested in real-world conditions, not just in controlled demonstrations.
Then, evaluate the fleet’s geographic distribution. Starship operates in both the US and Europe. This is relevant because regulatory environments differ significantly between regions. If you are based in Europe, you want a provider that has navigated European rules on sidewalk use, data privacy, and public liability. If you are in the US, you want evidence of deployments in states or municipalities with similar regulatory climates. The source material does not specify which cities or campuses Starship serves, so you should ask for a list of active deployment sites and, if possible, speak to existing customers.
Battery life and charging infrastructure are next. The source material notes that advances in electric batteries allow these bots to operate for tens of kilometers per charge. That is a useful baseline, but you need to know how that translates to your specific route lengths and delivery density. A robot that can cover 10 kilometers on a flat campus may perform differently on hilly urban terrain. Ask for range data under various load conditions and weather scenarios. Also, inquire about charging logistics—how long does it take to recharge, and where are charging stations located? The source material does not disclose these details, so you will need to request them directly.
Connectivity is another practical consideration. The source material mentions that upgrades in 5G and V2X networks improve remote connectivity and data collection. Before signing a contract, verify that your deployment area has adequate network coverage. A robot that loses connectivity in a tunnel or a dense urban canyon may need to rely on onboard processing, which could affect its ability to request remote help when needed. Walk your proposed routes with the provider and test connectivity at various points.
Remote assistance is a feature you should understand thoroughly. Level-4 autonomy means the robot can operate independently under specific conditions but may request human help in rare cases. Ask the provider how their remote assistance system works. How many robots can a single remote operator monitor? What is the average response time when a robot requests help? The source material does not provide these figures, so you must obtain them from the provider. Do not accept vague assurances; ask for documented response times and escalation procedures.
Finally, plan for integration with your existing operations. If you are a university or a grocery chain, the delivery robots need to interface with your order management, inventory, and customer communication systems. The source material does not detail Starship’s integration capabilities, so you will need to ask about APIs, middleware, and any customization required. A robot that cannot integrate smoothly with your backend will create more problems than it solves.
Common mistakes to avoid
One of the most common mistakes is assuming that a delivery milestone translates directly to your specific use case. Starship’s 8 million deliveries and 10 million miles are impressive, but they were accumulated across a variety of environments, customer types, and delivery categories. If you are planning a small deployment in a single neighborhood, the provider’s global numbers do not guarantee success in your context. You need to ask for data from deployments that resemble yours in scale, geography, and delivery mix.
Another mistake is overlooking the operational design domain. Level-4 autonomy sounds reassuring, but it comes with conditions. The source material explicitly states that these systems operate under specific conditions, such as mapped areas and good weather. If you deploy in a region with heavy snow, extreme heat, or frequent rain, you need to know how the robots perform under those conditions. Do not assume that because a robot works in California, it will work in Scandinavia. Ask for performance data under adverse weather and lighting conditions.
A third mistake is ignoring the human element. The source material notes that robots may request remote human help in rare cases. That means you need a plan for how those requests are handled. Who monitors the robots? What happens if a robot gets stuck and no operator is available? The source material does not specify staffing requirements, so you must clarify this with the provider. Underestimating the need for remote monitoring and intervention is a common pitfall that leads to poor service levels and frustrated customers.
Another frequent error is failing to account for regulatory variability. Starship operates in both the US and Europe, but the regulatory landscape for sidewalk robots is far from uniform. Some municipalities have embraced them; others have imposed restrictions on speed, weight, or operating hours. Before committing to a deployment, research the local rules in your target area. The source material does not provide regulatory details, so you will need to conduct your own due diligence. A provider that operates successfully in one city may face entirely different constraints in another.
Cost modeling is another area where mistakes are common. The source material does not disclose pricing, so you will need to build your own cost model. Include the obvious items—robot purchase or lease costs, maintenance, charging infrastructure, and remote monitoring staffing—but also consider less obvious costs such as insurance, liability coverage, and potential fines for regulatory noncompliance. Do not assume that the provider’s published figures on CO₂ savings or delivery volumes translate into a favorable return on investment. Run your own numbers based on your specific delivery volumes and operating costs.
A related mistake is neglecting to benchmark against alternatives. The source material frames Starship as a leader in the sidewalk-bot segment, but it also notes that other companies are active in last-mile delivery. Before committing to any provider, compare multiple options. Look at their fleet sizes, delivery volumes, geographic coverage, and technological approaches. The source material mentions that most delivery bots and robotaxis today are level-4 autonomous vehicles, so the technology is not unique to Starship. Your decision should be based on fit with your use case, not just brand recognition.
Another common error is underestimating the importance of data. The source material highlights that upgrades in 5G and V2X networks improve data collection. That data is valuable not just for monitoring but for optimizing routes, predicting maintenance, and improving service over time. Before signing a contract, ask who owns the data generated by the robots. Can you access it? Can you export it? The source material does not address data ownership, so you must clarify this in your negotiations. A provider that hoards operational data will limit your ability to improve your own service.
Finally, avoid the mistake of treating the deployment as a one-time project rather than an ongoing operation. The source material describes a system that has been running for years, accumulating millions of deliveries and crossing roads and driveways at a rate of two per second. That level of operational maturity comes from continuous monitoring, maintenance, and iteration. If you are not prepared to invest in ongoing oversight, your deployment will likely underperform. Ask the provider about their maintenance schedules, spare parts availability, and software update policies. The source material does not disclose spare-part lead times or response times, so you must obtain these commitments in writing.
In summary, the Starship milestone is a valuable data point, but it is not a guarantee of success for your own deployment. Look for evidence of operational maturity, understand the technical and regulatory constraints, and build a realistic cost model. Avoid the common mistakes of overgeneralizing from the provider’s global numbers, underestimating the need for remote assistance, and neglecting local regulations. By taking a disciplined approach to evaluation and implementation, you can increase the likelihood that your sidewalk delivery deployment will deliver the efficiency and sustainability benefits that the technology promises.
Sources
Published by Vigla Media OÜ (Estonia).