EXAMPLE THREE - Overcoming difficult spots and tricky situations on complex lawns
This article includes real-world examples of situations where Navimow's unique VisionFence camera really comes into its own. VisionFence brings powerful object recognition AI (artificial intelligence) capabilities to Navimow enabling it to conquer areas where other robot mowers can't reach.
First, some background. The Segway Navimow employs a range of strategies to help it get unstuck. For example, if it discovers it has a wheel is spinning, perhaps because a wheel has slipped off an edge or into a garden garden bed, Navimow tries a number of manoeuvres to climb out of trouble. These involve rotating each wheel in a mixture of directions and speeds, and trying to orientate the body of the mower in a direction where it can drive out of where it has unexpectedly found itself.
Navimow owners are most likely to encounter this kind of issue during their first runs after initial set-up. The solution is to edit the map so Navimow's map boundary does not run quite so close to an edge or other problem spot. But if Navimow has driven into a divot or rabbit hole that wasn't there yesterday, you might have to come to the rescue and fill in the hole. For example, here is a photo of where Navimow became bellied in very spongy kikuyu grass. Hidden underneath the grass mat was a deep depression in the soil where a horse hoof had landed when galloping across the lawn, forming a "hidden, covered pit trap". One of Navimow's wheels sunk into the hole, and while it tried every trick in the book to escape, in the end human intervention was required (of course, this is not the kind of situation many Navimow's will happen upon).
Navimow knows where it with an accuracy of 1 to 2 cm by making use of the known location of the EFLS transmitter (usually this is on 1m high pole close to the Base Station, but it can also be mounted on higher pole, fence, or even up on a roof). In conjunction with an ever-changing selection of GPS/GNSS satellites passing above, Navimow triangulates its position. Typically, there are about a dozen in view at any one time, with about half of them providing strong signals.
Navimow also has an inertial idea of where it is now versus where it was a moment ago, because it knows how its orientation and movement at any given moment: detecting yaw, pitch, roll, and linear acceleration using sensors sensors, plus it knows the rotational speed of its wheels (and therefore the distance it has travelled since the last "known" point). This completes the Real Time Kinematic Positioning System.
If the signal from the EFLS unit is weak, and/or signals from GPS/GNSS satellites are obstructed, and/or the satellites overhead are minimal in number then a robot lawnmower can lose track of exactly where it is. It employs recovery strategies, including pausing and waiting for more satellites to come into view, and moving around a little in the hope of finding a spot with improved reception. If it cannot recover its location, it stops and sends a message to its owner asking to be carried to a more open space and restarted so that it can carry on working from where it left off. Unless, of course, it is a Segway Navimow with VisionFence turned on.
Our experience with running Navimow in a typical Auckland suburban property has revealed that Navimow can operate successfully even when the EFLS transmitter is mounted on a 1m pole and the signal needs to travel through the entire width of the house to reach the lawn on the other side.
In the photo below (Figure 1), the signal travels along the green line, through the house and garage onto the most obscured area of Zone 6 without causing problems.
More impressively, the signal has also been found to easily travel along the yellow line, passing through the entire house, to the most obscured area of Zone 7 without causing problems.
However, while working in Zone 7 (with VisionFence turned off), occasionally Navimow will pause from time to time while it waits for enough satellites to come into view. Zone 7 is shaded by a very steep bank that blocks out ~30 degrees of the skyline in a southerly direction, and it bounded in the other three directions by roofs. To date Navimow has always recovered after a brief period, and continues on the job without human intervention. Turning on VisionFence enables Navimow to operate even when satellite signals are attenuated or obscured. It is even possible to mark areas of your lawn where VisionFence is to be used and where it can be turned off. Remember, VisionFence cannot "see" at night, so you can select whether or not you want Navimow to operate at night in a particular area where you've decided VisionFence brings benefits. That is, you can select if you want Navimow to operate 24 hours per day, or only during Daylight. Equally, if VisionFence is ON then you can select if you want Navimow to stop mowing at Night, or continue mowing at Night but without the benefit of VisionFence providing added situation awareness over and above Navimow's EFLS, RTKP, GPS/GNSS and touch sensors.
Every property is different, and signal attenuation will depend on the construction material of the house (e.g. concrete absorbs more signal than wood), number of walls the signal needs to pass through, presence of windows, etc (in this respect it is very much like home WiFi transmission). Our testing suggests Navimow will perform admirably on the vast majority of Kiwi suburban sections, even with VisionFence turned off. Turning it on brings added benefits no other robot lawnmower can offer.
The maximum range of the EFLS transmitter in an open space is ~170m.
At our rural property, the furtherest extent we've tested is 85m, with a thick 2m hedge in between the transmitter and the far corner of Navimow's map. In Figure 2 below the Yellow line is 85m. As expected, no reception issues have been encountered at this distance.
The Green line shows how the signal must pass through some ~30m of house to reach the most obstructed area of lawn. Again, no reception issues have been encountered despite the extent of this impediment.
Equally, Navimow encounters no troubles operating under tall, established trees planted on lawns and boundaries.
The Pink line in Figure 2 is a distance of about 35m between the EFLS transmitter and the boundary, so the signal at that point is still very strong. However, the area in the Pink oval is the spot where Navimow has got itself into a little bit of trouble on two occasions over a period of 4 weeks mowing. As you can see in the photo below, Navimow has gone off the beaten track!
Let's learn why this happened despite a strong signal from the EFLS transmitter, and how VisionFence solves this problem.
This area of the lawn is a "canyon of trees" formed by two rows of tall trees - a cypress shelter belt and three specimen trees surrounded by a planted garden bed. There is just a thin strip of grass to mow in-between, and a narrow strip of blue sky above (see pink oval in Figure 2). Here some additional photos for further context.
Most of the time Navimow is able to mows this "canyon" without any problems, because the EFLS signal is strong, the inertial guidance systems are clever, and there are enough satellites above with a sufficiently unattenuated view to the sky. It is not that trees entirely block the signals from the satellites, just attenuate the signal to varying degrees depending upon foliage, atmospheric conditions, and the position of each satellite in the sky.
On two occasions over 4 weeks Navimow has got a bit lost in this "canyon". He's stopped, called for help (via sending a message in the App) and asked to be moved to a more open space and set going again. Figure 3 captures the one occasion when he'd driven himself into the tree line while attempting to find a spot from which to re-establish contact. Both times Navimow got "lost" we had VisionFence turned off. After turning VisionFence on, Navimow has never again got "lost" in the canyon of trees because it is able to continue to operate safely by relying on real-time AI visual analysis of the operating environment.
VisionFence is a game-changer for enabling Navimow to operate in uncommon situations.
All Navimow models sold in New Zealand feature VisionFence.
In Figure 2 there are two Light Blue ovals marked out on the satellite property map above. These are areas of Zone #2 that get very wet in the winter because they are part of the drainage system. These areas will be marked off as temporary no-go zones during July and August - a quick and easy thing to do in just a couple of minutes by using the Navimow App. When the ground dries out again these temporary exclusion zones can be deleted.
Our real-world testing has shown that on almost every suitable property Navimow will work day in, day out, keeping your lawns looking perfect all of the time without any human intervention at all. Just set up your schedules and Navimow will do the rest. If you have a particularly complex or unusual environment you may need to give your Navimow a helping hand once in a while if you have VisionFence turned off.
Call 0800 SEGWAY for more information, pricing and availability for Segway Navimow.