Delivery robot company Refraction AI closed $4.2M in seed funding this week. The seed round was led by Boston-based Pillar VC, and includes Osage Venture Partners, Invest Michigan, eLab Ventures, Truck Ventures Capital, Alumni Ventures Group and Chad Laurans. Refraction AI is currently beta testing its autonomous delivery solution, called REV-1, in the Ann Arbor, Michigan area. The company is working with local restaurants in Ann Arbor to deliver food orders to local residents.
To date, the company has raised a total of $10.7M over three rounds. Last-mile delivery is an automation market that has had several starts and stops since 2017, when delivery robots were banned from public streets in San Francisco. Since that time, the companies have worked with local governments to enable the testing of autonomous delivery robots on public streets and sidewalks.
Companies like Refraction AI continue to work with local city governments to determine the best way to roll out the services while minimizing the impact to local communities. Most delivery companies have focused on deploying the solution on college campuses, where the robots deliver food to students on campus. This use case has enabled the evolution and testing of guidance and navigation systems in a controlled environment, and with a willing and compliant partner (i.e. the university).
There are several ground-based competitors in the last-mile delivery space, including Kiwibot, Starship Technologies and Postmates. All of these companies are working through the issues of safely navigating sidewalks and streets, avoiding pedestrian and vehicle traffic, crossing streets at intersections and tracking delivery operations. In general, take-out operations for restaurants have increased during the pandemic, opening up the market for food delivery to customers who may not have considered this option prior to the pandemic. The Robot Report Podcast featured Kiwibot in an episode on delivery robots. Starship Technologies also reported earlier this year that they have passed the 1 millionth delivery milestone.
The market for urban food delivery is the low-hanging fruit that all of these companies are chasing. The urban environment is one of the more controlled and uniform operating environments for the machines. It will be a while before any of these solutions can be implemented for rural delivery, or even longer distance operations. It’s likely that a different form factor will be necessary in the types of autonomous vehicles deployed.
Refraction AI was launched in 2019 by University of Michigan professors Matt Johnson-Roberson and Ram Vasudevan. The company is based in Ann Arbor, MI and continues to work with local restaurants and grocery stores, exclusively for food delivery on campus and nearby neighborhoods. Unlike some of its competitors which are restricted to sidewalks, REV-1 operates on the street and is classified as an electric bike (ebike). The machine perception onboard REV-1 uses 12 cameras to see the world around it. It also uses radar and ultrasonic sensors to increase its awareness. The company had purposefully avoided the use of expensive LiDAR sensors to help keep costs down in the solution.
Being based in Michigan means Refraction AI has been able to engineer and test the system in rain and snow. Inclement weather operations have, until now, been a weakness of autonomous outdoor operations. As the video below illustrates, this technology has to operate in all environmental conditions if it’s going to truly become a viable option.
The market has never been friendlier for autonomous delivery robots. The pandemic accelerated the growth of e-commerce, and social distancing during the pandemic made pick-up, dine out and food delivery a viable option for many consumers who might not have considered these options during normal times. Refraction AI is one of several companies that are bringing real solutions to market and working to expand their client base. With the recent funding announcement, Refraction AI gains the necessary investment to position them for growth and success going forward.
Editors Note: This article was originally posted on The Mobile Robot Guide.