How M.52 Sees the World

The Importance of Vision 

Think about how you use your senses to move around a property while mowing. You’ll hear the motor revving up and the sound of the blades cutting grass. You might feel a deep hole or unexpected rut and avoid it next time. But most importantly, you use your vision to see and safely mow around the people, pets, and other obstacles you come across. Without the power of sight, safely and productively mowing a property would be impossible. 

Robots of all kinds are equipped with different instruments that help them sense their environment in similar ways. Ultrasonic sensors use sound waves to identify obstacles. Though they are at frequencies we can’t hear, robots with these sensors interpret the sound waves to safely operate on their own. Some robots have bump sensors that feel when they run into something and direct them away from the obstacle. This is how robotic vacuums find their way around the house (but it’s not a good approach for a commercial mower, since you don’t want it running into things at all). 

Just like sight is the safest and most important sense for a human to use when mowing, computer vision is the primary way Scythe M.52 navigates around a property. This technology, which we call Scythe Sight, enables the mower to see and respond to its surroundings in the same way a human operator would. And while M.52 uses a whole suite of complementary sensors, computer vision ensures it is mowing with enhanced safety and productivity. 

Scythe Sight: How Computer Vision Works 

Computer vision works like human vision, humans just have a head start with a lifetime of training through sight. With our eyes, we learn how to tell objects apart and categorize what those different objects are. We then learn what to do when we come across those objects. It might seem like second nature for a landscaper to mow around a tree on a property, but that basic action is built on years and years of learning what trees look like and what to do when mowing near one. 

Scythe Sight, M.52’s computer vision, uses rich visual data from the mower’s eight cameras to understand the world around it. We’ve shown the robot tens of thousands of images with people, animals, sign posts, fire hydrants, and lots of grass and then taught it what to do when it comes across each. As a result of this process, commonly called machine learning, M.52 is able to detect an obstacle, identify it, and take the proper action–all on its own. 

• M.52 sees and moves around static obstacles like trees and poles with no need for the operator to trace them
• M.52 stops and turns the blades off for dynamic obstacles like humans or animals, starting up again when its path is clear
• M.52 stops and notifies a nearby crew member when it runs into an obstacle it does not know, asking them if the obstacle needs to be moved or if it can mow around it 

Scythe M.52 sees more with every mow, meaning its ability to identify and respond to obstacles will increase significantly over time. As far as we know, Scythe already has the largest data set of landscaping scenes there is, and this growing data set enables M.52 to mow more on its own. 

20/20 Vision 

You wouldn’t want someone with poor eyesight driving a mower without their glasses, and the same goes for robots. Because Scythe Sight is so important for how M.52 operates on its own, we wanted to give it the best set of glasses we could find.

Eight High Dynamic Range cameras provide M.52 with 360º of the richest visual data possible for the mower to analyze as it mows a property. Compared to other systems, HDR cameras provide M.52 with more detailed imagery to use as it navigates the area, making for better autonomy. HDR cameras also help M.52 see in greater detail across a range of different and difficult lighting conditions, enabling the mower to perform better in shifting shadows or sun glares than other kinds of cameras. 

An Eye for Opportunity 

As Scythe M.52 learns more about mowing environments, it will not only be able to see and respond to obstacles but it will also be able to identify upsells. Scythe Sight will spot broken sprinkler heads, identify brown spots in turf that need treatment, and flag other service opportunities for your account managers to pitch to your clients. 

See the Future

Scythe Sight enables better autonomous mowing, which expands the capacity and performance of your crews. They will spend less time managing the mower and more time doing higher-value work on site. To learn more about how the computer vision powering M.52 can give your business futuristic foresight, check out our technology page. 

The Scythe Approach

Offering a deep dive into its design, this is the third in a four-part series explaining the pillars of our approach in developing Scythe M.52. Together these choices result in the most comprehensive autonomous mowing solution with better performance, greater durability, and increased productivity for landscaping crews.