As a recent study indicated, aerial delivery drones aren't always as energy-efficient as ground-based transport. We may therefore be seeing more wheeled delivery robots, such as the rather cool-looking and increasingly in-demand REV-1.
Somewhat similar in appearance to the Go-One 3 velomobile, the REV-1 was developed by University of Michigan spinoff company Refraction AI.
It's currently the focus of an ongoing pilot project in the university's home city of Ann Arbor. Starting late last year, a small fleet of the robotic vehicles began delivering food from five local restaurants, as summoned by approximately 500 customers via a custom app.
Since the COVID-19 crisis began, though, there has reportedly been about a four-fold increase in use of the service. A total of eight of the wheeled robots are presently in use – they now have ultraviolet sterilizing lights inside their food compartments, plus they're wiped down between each delivery.
With a top speed of 12 mph (19 km/h), the REV-1 weighs 100 lb (45 kg) and is powered by a less than 500-W motor. This means it's low-powered enough to qualify under e-bike regulations, so it generally sticks to bicycle lanes. That said, it can still nip out into car lanes as needed, such as when making left-hand turns.
The robot navigates the city and avoids obstacles autonomously, using a combination of GPS, LiDAR, radar, and 12 optical cameras – some of the latter have a wide field of view of around 200 degrees, while others have a more narrow field of view of around 90 to 100 degrees. This spread of cameras gives the REV-1 a full 360-degree view of its surroundings, and enables depth perception. And because it travels relatively slowly, it is also able to make use of ultrasound sensors, which aren't useful in faster-traveling vehicles.
Additionally – given that it was designed in Michigan, which gets cold winters – the vehicle has a fat rear tire to help it power through snow.
It is hoped that a total of 25 REV-1s will be on Ann Arbor streets by the middle of the summer. You can see one in action, in the video below.
Sources: Refraction AI, University of Michigan via IEEE Spectrum
