Delivery robots are on the fast-track to the mainstream.

Companies including UberDominos, and Grubhub are currently testing the technology, while the artificial intelligence startup Starship has delivery robots zipping around college campuses across the U.S. The potential benefit for business owners with brick-and-mortar stores or restaurants who could make local deliveries of goods without hiring a delivery person is immense. Imagine if you owned a hardware store and someone called to ask if you have a specific product. You could simply put the item in a delivery robot and send it on its merry way.

Autonomous delivery via self-driving cars, on the other hand, is likely further off. Development of the technology has been a slow process, as the machines have continued to plague developers with crashes, sudden stops, and other unforeseen events on the road. These problems, referred to in the industry as "edge cases," are the main deterrent to the wider adoption of self-driving cars, according to Ali Kashani, founder of delivery robotics company Serve, which was spun off from Postmates' autonomous delivery division in 2021.

"Imagine you're a self-driving car, and your cameras and sensors register that something is in front of you. You might not be able to tell if it's a shadow or a real object," Kashani says. "You have moments to choose between two options: keep going or stop. Both options pose a huge safety risk." Serve makes shopping-cart-shaped robots that deliver food in urban environments by traveling along the sidewalk rather than the road. Currently, the robots are being tested by Uber Eats and will be rolled out for Chili's. 

While A.I. hasn't advanced enough to not pose a risk to drivers, the technology is always advancing. In July 2022, a group of researchers at Google's Deepmind program published a study stating that they had developed an A.I. with a basic understanding of physics, akin to a human baby's conception of gravity, object permanence, and directional inertia. This kind of research, according to Kashani, is likely to solve the self-driving car problem, but it could take decades.

Instead of waiting for the technology to improve, Kashani suggests that entrepreneurs and business owners look for solutions where A.I. has a defined application and an immediate benefit -- where the difficulty of the task is equal to or lesser than the algorithm's abilities.

"We are simply not ready for the road, but the problem is easily solved when we put the robots on the sidewalk," Kashani says. The major factors that make sidewalk robots significantly less risky than cars is that they're smaller, lighter, and slower, so they have a very small risk of causing injury even if a mistake is made. Sidewalk robots also have the ability to stop for a moment if they need additional time to process, a luxury cars don't have.

Delivery robots could also be a boon for entrepreneurs looking to cut down on their emissions. Kashani says that roughly 1.5 percent of all CO2 emissions come from cars making shopping trips within a three-mile radius of their homes. Sidewalk robots are much less energy intensive, with Serve's robots using 85 percent less energy than a two-ton car.

There are some parts of the delivery experience that sidewalk robots simply can't replicate, however, such as entering an apartment building and scaling stairs, but Kashani believes people will be more eager to run downstairs if they don't have to tip a driver. While it may be a while before you can send a car out to deliver a burrito, smaller autonomous vehicles are making strides that are expected to bring much-needed innovation to business owners.