Since the first Roomba came out in 2002, it has seemed inevitable that one day iRobot would develop a robotic lawn mower. After all, a robot mower is basically just a Roomba that works outside, right? Of course, it’s not nearly that simple, as iRobot has spent the last decade or so discovering, but they’ve finally managed to pull it off.
Today, iRobot is previewing its Terra robotic mower. It’s rugged, fully autonomous, and literally wireless, using radio beacons to localize rather than relying on a buried edge wire to keep it from mulching your begonias. We don’t have all the details yet—including price and availability—since iRobot is being a little bit coy, but we do have pictures, a video, and a chat with iRobot CEO Colin Angle about why developing Terra took them so long.
While this announcement is certainly a surprise, we’ve had a pretty good idea that iRobot was getting serious about a robotic mower since at least April of 2015, when we spotted this FCC application for a robot mower from iRobot that used wireless beacons to navigate. The fact that iRobot was willing to take on angry radio astronomers in front of the FCC suggested that they were serious about getting this to work, and thankfully, the issue was resolved (more on that in a bit). At the time, we’d been optimistic about this method of navigation, since it seemed to solve one of the major problems that robot mowers have struggled with for years. From our 2015 article:
Just about every single robotic lawn mower that currently exists (and there’s a bunch of them) requires you to first bury a loop of wire called an edge wire around the perimeter of your lawn to keep the robot contained. The robot senses the wire under the ground and won’t cross it. This is a pain to set up, and historically it’s not the sort of thing that iRobot wants to ask users of its robots to do.
iRobot’s approach, according to a waiver that they’ve filed with the FCC, is to replace that edge wire with a series of small wireless beacons… After an initial user setup, [iRobot’s mower] would use the beacons to calculate its position using time-of-flight of transmitted ranging packets.
Plant the beacons around your yard and Terra will localize itself—no perimeter wires required. Photo: iRobot
This localization and navigation method is much different from the way that modern Roombas navigate. The visual SLAM method that Roombas use indoors wouldn’t be reliable outside, because the navigation cameras would likely get dirty, and the sun would cause all sorts of problems. GPS isn’t nearly accurate enough, and satellites can get occluded by trees anyway, so a wireless solution was ideal. iRobot isn’t sharing the sort of nerdy technical details that we’re really interested in, like how accurately Terra can localize, but the video makes it look quite good.
The setup process is slightly more complicated than unleashing a Roomba for the first time, but not by much. You plant the beacons around your yard, and then drive the robot around the perimeter of the area that you want it to mow using an included joystick. You also drive it around any areas inside that area that you’d like it avoid, like flower beds or fishponds. And then you’re done, and it’ll mow those areas autonomously in a nice, neat straight line pattern, charging when it needs to and then returning to where it left off. The beacons themselves are battery powered and should last about a year, and iRobot expects that you’ll just swap the batteries when you bring everything in for the winter.
After driving Terra around the perimeter of the area that you want it to mow using an included joystick, you can set the robot to mow those areas autonomously. Images: iRobot
One of the advantages of an autonomous mower is that you can have it mow more frequently than you would, meaning that a mower that “mulches” grass is realistically useful. The idea behind mulching is to cut grass into pieces that are small enough to get turned into fertilizer, but the pieces have to be quite small for that to work well. You can buy fancy mowers that will attempt to puree grass of any size, but iRobot figured that with Terra, you’ll just have it take a little bit off the top once a week, or even more often, because why not? Your lawn won’t just look better, but it’ll be healthier, too.
From the sound of things, it’ll be a little while before we get to check Terra out in person. It’ll officially launch in Germany sometime around the beginning of Q3, and there will be a limited beta program here in the United States, followed by an official launch “at a later date.” In the meantime, we were able to ask iRobot CEO Colin Angle about what he’s going to do now that they don’t have a secret robot mower to work on.
IEEE Spectrum: So, a robot mower, finally!
Colin Angle: This is probably the most requested robot in iRobot’s history after the Roomba—probably two minutes after announcing Roomba, people were like, “When are you going to mow my lawn?” It’s turned out to be a much more challenging robot to build, and we’ve finally gotten to the point where we’re announcing it.
The goal was, first off, to do a great job cutting, and to mow the lawn in a way that people would find easy to set up, and would work in the complicated lawns that many homeowners have. So, no boundary wire was a critical design element.
The fundamental navigation system was a technology that we had to develop, and it was frustratingly challenging. We found a workable solution using ultrawideband radio direction finding technology. You put in a couple beacons, you take the robot out of the box, you drive it around the perimeter of your lawn and the interior features, and you’re done. And we thought that was a good customer experience, and it allows for the sort of physical programming that made the original virtual walls on Roomba quite successful.
And then we had to go build a robot that was durable enough to operate and survive outside—we have a lot of experience there, from building military robots. And we had to design a cutting system which combined safety and effectiveness—cutting a blade of grass correctly is actually really hard. We invented a novel self-retracting blade that balances the need for safety with the mission of giving a cut comparable to the type of cut a high quality mulching mower would give. It’s a really cool solution, and we believe that with Terra, you buy this robot and you’ll never mow again.
How long was the development process for Terra?
Truth be told, we probably started this project in the 2005 timeframe. There was a little bit of a hiatus while Roomba required all of our attention, and then this was relaunched about five years ago, based on the work done before. It has been a long time coming, and people have wanted it for a long time. I’m super excited to not have to answer any more questions like, “So when are you going to have a lawn mower?”
How is developing a robot lawn mower different than developing an indoor vacuuming robot?
I would say that the performance bar for version 1.0 for a robotic lawn mower is very different from the first Roomba. With Roomba, when we first launched it, we didn’t claim that it deep cleaned like an upright vacuum. This was a routine cleaner; you could use Roomba every day, and then once every couple weeks, pull out your upright and use it. It wasn’t until the 900 series where we really got to upright vacuum levels of performance. For the lawn mower, you can’t kinda sorta mow—you’re either mowing, or you’re not mowing. So one of the challenges was getting up to that replace your traditional push mower functionality at launch, which made the whole thing harder.
How has iRobot’s experience with building robots for the military impacted Terra’s development?
Building something for the outdoors and that can operate in rough terrain is definitely a core competency that was developed within iRobot, and it certainly did not leave when Endeavor was spun out. So the idea of, how do you handle sealing, how do you handle shock, how do you handle power levels on your robot that are much higher than what a Roomba would typically endure—the infrastructure that we have for environmental, shock, and vibration testing are things that Terra benefited from. Even thinking through, what are reasonable specifications for operating conditions, which are broader than you might think, if you weren’t experienced with developing outdoor robotics—that proved to be very important and beneficial.
How much variability have you found in people’s lawns as compared to rooms in people’s homes? Is it more complicated, or less complicated outside?
It’s different complicated. You have hills and grades and valleys and different types of grass that are easier or harder to cut, and different hazards than are in the home. The hazard that gave us a lot of problems most recently was the outside trampoline, which was the equivalent of Roombas and Ikea wire chairs, which have a base with a thin inverted “U” shape which drove us insane for a while with Roomba. We went out [with Terra] and were feeling good about our first alpha tests, and then discovered the trampoline, with its stupid “U”-shaped legs. But we’re feeling much better about this nemesis of the outdoor lawn mower these days.
As with Roomba, users can program and monitor their Terra mower using an app. Photo: iRobot
What kind of homeowner did you have in mind when you designed Terra?
If you hire someone to mow the lawn for you, this is going to be an attractive alternative. If you mow the lawn yourself but would prefer not to, or you mow the lawn yourself and realistically don’t always get to it, this is going to be great. Like vacuuming, there are actually people who love mowing—they view it as a chance to do something outside when no one can talk to them, and they can stay mowing. But for the rest of us, this is an accessible alternative to paying someone to do it for you, or a more reliable way of getting that particular task getting ticked off your to-do list.
Can Terra handle noncontiguous areas of lawn?
Terra is a platform, which means that it will grow in its capabilities over time. At launch, places need to be contiguous, but it’s definitely something that the robot is very capable of doing. The hardware is designed so that you could do your front lawn, show it a path to get to the back yard, and then expect it to get to the back yard—the whole system is capable of that. Whether that feature will be a launch feature or not… Probably not in 2019, but by the time we’re scaling things up in 2020, certainly a reasonable thing to imagine.
A couple years ago, iRobot filed an FCC application for the wireless beacons for what we now know is Terra, and some radio astronomers were worried that the beacons might cause problems for them. What ended up happening with that?
I can affirmatively state that no radio astronomy equipment is going to be rendered inoperable by a proliferation of Terras. Based on the power of our ultrawideband beacons, we were able to insure that we were not going to interfere. I think that if you’re a radio astronomer, any amount of energy is bad, including cell phones, but that one got far more attention than it probably deserved. We’re good, and SETI is intact.
We’ve talked before about how much potential there is for home robots like the Roomba to integrate with and improve smart homes and IoT tech. Does an intelligent outdoor robot like Terra fit in with that?
Absolutely. It’s a connected product, and you can imagine wanting your lawn mower to be capable of talking to your irrigation system, or being weather-aware. It’s definitely extending the power of what a home ecosystem can do for you, into the yard. Maintaining your home requires a system, and Terra is fully integrated into our vision of how this is going to work out.
Now that you’ve made a lawn mower, can you speculate a little bit about what other opportunities there might be for affordable, practical home robots in the near term?
I’ll give you a directional answer—with the Roomba i7, we’ve opened up a whole new arena of things that robots can do, because we’re starting to understand where stuff is, and what the home layout is, so that traveling from Point A to Point B is possible. If a robot need to go to the kitchen to get a beer, we at least now know where the kitchen is. So there’s this idea that two-dimensional surface cleaning robots, including Roomba, Braava, and now Terra, have now opened themselves up to a more destination-oriented world, where robot arms suddenly make sense. I think as we look forward, we’re going to see a lot more opportunities to build a whole new level of functionality into the robots that we create—like carrying things.
So you’re optimistic about mobile manipulators in the home?
Oh absolutely! We’ve done stuff in the past—we have R&D projects on low-cost robot arms. But if you don’t know where anything is, it’s really hard to justify the cost. Picking things up is about moving them, and if you don’t know where to move them to, the whole value proposition breaks down, and that’s now changed. And that’s a big change.
Colin Angle has been consistently saying for several years now that we should be looking forward to in-home robots that can move objects, and it’s likely the case that iRobot itself has more in-home robotics experience than any other robotics company. It’s interesting to consider how they might take a (literally) ground-up approach to a mobile manipulator, building on their experience with Roomba, or perhaps Ava, and seeing where it goes from there.
For now, though, we’re quite happy with Terra to look forward to. As soon as iRobot makes more details public, we’ll bring them to you, and we’re also hoping to get our hands on a review unit this year—as long as it comes with a nice big yard with a trampoline in it for us to try it out on.
[ iRobot Terra ]