Two University of Michigan Engineering professors have developed a robotic food delivery service that transports food from local restaurants to customers’ homes. Engineering professors Matthew Johnson-Roberson and Ram Vasudevan founded Refraction AI in May 2018, the company which designed the new autonomous delivery vehicle, named REV-1.
The REV-1 is a compact and lightweight vehicle designed to operate in both car and bike lanes, maximizing safety due to its small size. The vehicle uses 12 cameras, radar and ultrasound sensors to navigate and look out for possible obstructions while on the road — including potholes, trees, buildings and poor weather conditions.
Johnson-Roberson said because of challenging road conditions in Michigan, Ann Arbor was a good place to launch an adaptable and safely-operating robot.
“Ann Arbor is a really great location for a couple of reasons,” Johnson-Roberson said. “A lot of autonomous vehicle deployment is happening in Arizona and California where the weather is the same every day and the roads are really wide and very flat and have no potholes, so it’s not terribly representative for the rest of the United States. Most of the United States is not like Arizona, and so in that way I think this is a great place to deploy robots because you have to deal with a lot of the things that other people are just sort of avoiding, and I think it’s really forcing us to make a better robot.”
Johnson-Roberson said Ann Arbor was also an ideal location to launch the new autonomous food delivery service due to its high restaurant density.
“These robots are really designed to go between a half a mile and two and a half miles— maybe three miles at max and so there’s a ton of density of restaurants (in this range),” Johnson-Roberson said. “If you think about how many restaurants are downtown, we can pick up from probably 40 places before we even have to go more than three or four blocks.”
Johnson-Roberson said he hopes to eventually expand the service into cities outside of Ann Arbor, particularly Boston and other college towns like Madison, Wisconsin.
During the pilot program, Refraction AI partnered with local restaurants including Miss Kim and Belly Deli. In the future, they hope to expand to work with restaurants all across campus, as well as collaborate with MDining to open up a delivery service on North Campus and execute delivery from their dining service Fireside Café.
Eric Joh, co-owner and operational manager of Belly Deli, said the restaurant was excited to partner with Refraction AI and believes the partnership will generate public attention for the restaurant.
“We thought it was a neat idea, and I think probably the future of deliveries will be automated,” Joh said. “We thought it was pretty exciting what they were doing at Refraction AI, so we wanted to be a part of it. I think it’s going to increase business by generating a lot of publicity; I think it generates a lot of public interest and it’s something very exciting.”
Joh said the cost was one of the advantages of this service over the traditional delivery services they use like Postmates and Grubhub.
“If cost is an issue, people wouldn’t have to tip as much for an automated service — it would already be included in pricing.” Joh said. “As of now it is a little bit cheaper than the other services. The commission rate is slightly lower than the other delivery services we use. If they have enough vehicles on the road then I think most of the companies will use it because it’s more cost effective, but I think that’s way further down the road.”
Johnson-Roberson has spent his career working on self-driving cars and other robots and said he wanted to create a robot that regularly interacted with humans and improved their lives.
“I was disappointed that there’s still not a robot that touches your life everyday.” he said. “You go to school, you go home — you don’t see a robot. Maybe you occasionally see a Waymo car but it’s not doing anything for your life. It’s been my career building robots, and I want to build a robot that you get to see and for my neighbors and my friends… it makes their lives better and it helps to improve their day to day and interact with them regularly.”
LSA junior Blaine Thompson is a frequent user of food delivery services and says he thinks the new autonomous delivery service would be useful for Ann Arbor traffic and speed of delivery.
“Most of the time I look for the closest restaurants to avoid a big delivery fee, unless I want something from a restaurant on North Campus or something,” Thompson said. “I would use a robotic service like that mostly as just another delivery service, but having less cars on the road is always good for the environment and Ann Arbor driving. Likewise, it seems that it might be quicker than traditional food delivery services.”
The service had five robots during their pilot program, and the founders of Refraction AI anticipate to expand the fleet to between 15 and 30 robots into the winter and spring.
Johnson-Roberson listed three major factors that separate this compact autonomous delivery service from a traditional vehicle and delivery driver: safety, environment and reliability. He said the REV-1 is less dangerous than a traditional larger vehicle and in the event of an accident, customers can only expect your food to get damaged — not a human driver.
“We have the idea that if you go slower and make something that’s lighter it’s a lot safer… and you can do it without having to have the same safety requirements you would have for a bigger vehicle,” Johnson-Roberson said. “If you don’t have a person in it the safety requirements are so different because if it’s a burrito inside, you don’t really care if the burrito gets damaged if the vehicle gets hit.”
Each robot costs about $4,000 to build and travels at a steady speed of between 10 and 12 miles per hour, putting the robot at less risk of damage if an accident occurs.
Johnson-Roberson said another benefit of the robot is that it is more environmentally conscious than a traditional vehicle because it is electric and does not rely on fossil fuels.
“There’s really a lot of issues around sustainability for using cars to do food delivery, and so we thought one of the big ways that we can have a long-term impact is moving us away from using big fossil fuel-burning cars to do delivery when we can do it with an electric vehicle that’s as light as a bike,” Johnson-Roberson said.
Johnson-Roberson also mentioned the robot is more reliable than a traditional delivery driver, due to lack of concerns about an employee getting sick or having to miss work. Additionally, he said the robots are serving a market need, rather than taking away jobs from human delivery drivers who do not have sustainable jobs.
“We don’t think about that lightly and I think it is important to realize what the delivery market is like,” Johnson-Roberson said. “Deliverers don’t get health insurance, they don’t have any job stability, they’re dependent mostly on tips so it’s not a very sustainable job. We hope that we’re building something that’s a more sustainable industry that treats its workers more fairly and that hopefully that humans at the end of the day are going to be the ones that benefit from this.”
He said the tele-operator employees working for Refraction AI will receive benefits food delivery drivers typically don’t have.
“We’re not trying to exploit people to make this kind of business work,” Johnson-Roberson said. ‘We think that there is a way to pay people a fair wage to do things like remote operation, and we can do that in a way where they are stable employees — they get health insurance; they get all the benefits that are important. We think that this technology, and not just this, but many of this autonomous driver technology is going to make big changes in our economy, and we would like to make sure we do that in the most ethical and most conscientious way.”