Refraction AI entered into the $80 billion robotics market in 2019 with a mission to bring a practical, economical and deployable robot to the last-mile delivery space.
While working at the University of Michigan’s Robotics Program, co-founders Matthew Johnson-Roberson and Ram Vasudevan noticed a gap in autonomous vehicle (AV) mobility: extreme weather conditions. The pair developed their “Goldilocks of autonomous vehicles,” REV-1, a last-mile delivery robot that could handle the harsh weather a majority of the country experiences.
In developing REV-1, Refraction AI discovered many last-mile AVs used a sensor system known as LiDAR, which was causing bottlenecks in deploying these vehicles into the real world.
LiDAR, an acronym for light detection and ranging, determines ranges with a laser by measuring the time for the reflected light to return to the receiver. These sensors are not able to recognize different human behavior, like a cyclist looking over his or her shoulder before changing lanes. They can find it difficult to register winter weather if not used in tangent with other sensor systems.
These LiDAR- based sensor systems also can be incredibly expensive, making the investment in AVs challenging for businesses that are looking for last-mile solutions.
The REV-1 uses a combination of cameras, millimeter wave radar and ultrasound sensors to make up for the depth perception and mapping advantages of LiDAR sensors. Refraction AI says it has found its testing results just as accurate, and more importantly, deployable in real-world markets.
In 2020, a year after founding the company, Refraction AI launched its delivery robots in Ann Arbor, home of the University of Michigan.
In an interview with FreightWaves, Luke Schneider, the CEO of Refraction AI, explained why Ann Arbor was chosen to launch REV-1.
“Michigan’s harsh weather conditions give us a great control environment for REV-1,” Schneider said. “It also gives us the opportunity to test in a city environment with high-traffic areas.”
He said the REV-1’s size is important as it needs to be lightweight to control costs and provide safety while maneuvering around highly trafficked areas. The AV robot weighs about 80 pounds and is 5 feet tall and 4.5 feet long. The inside of the robot has enough space for about four to five grocery bags. These dimensions give the REV-1 the opportunity to take heavier loads than smaller AVs in the market and yet with a top speed of 15 miles per hour and a stopping distance of about 5 feet, qualify it under e-bike regulations.
“The REV-1 is designed for Michigan winters and snow on the road; this is why it has a third wheel,” said Schnieder while explaining the unique structure of the robot.
In early March, Refraction AI announced a $4.2 million seed round led by Pillar VC and including eLab Ventures, Osage Venture Partners, Trucks Venture Capital, Alumni Ventures Group, Chad Laurans and Invest Michigan. The company plans on using these funds to add more customers and deploy more robots in Ann Arbor.
Currently, a fleet of 25 robots are available to nine different restaurants, grocers and retailers throughout the city.
“Our goal is to add more capacity for Ann Arbor businesses,” explained Schneider. “We do plan on expanding into different markets, but our focus is to enable these companies with a different way to deliver to customers that keeps fleet costs low and is available to them now.”