Non-'Perimeter Wire' Robot Lawnmowers

[martinwinlow]

Hi,

Does anyone know of a robotic lawnmower that does not need a perimeter wire to be installed first? I was thinking of something like the Neato Botvac robotic vacuum cleaner which uses LIDAR (I think) to scan its working area and then works out the most efficient path to clean the room... LiOn batteries, please and wifi enabled would be ideal (I know I'm not asking much!)

Regards, Martin.

[glnc222]

Vacuum cleaners have interior walls to identify the area, besides other boundary markers used, magnetic strips and light beams or ultrasound. Not so with an outdoor lawn, not to mention flower beds etc. How you would expect a mower without boundary markers of some kind to work is not clear. Imbedded wires is the best way to signal the boundary outdoors, compared to light beams say, with the less controlled outdoor environment, large in size etc. Wires are also used indoors on industrial robots. Whatever else there might be would be very expensive, like a self driving car. I gather the Husqvarna supports three guide wires compared to one on Robomow, for more elaborate layouts, but check model specs.

[glnc222]

Even carbon based mowers like horses or goats require boundaries (like a fence).

[glnc222]

Note robot mowers can still do systematic travel within boundaries; the entire lawn is not filled with a guide wire track, as I understand them.

[martinwinlow]

Sure, *some* gardens don't have hard, vertical boundaries, but lots do. Mine for instance. Flower beds would/could look like the top of a flight of stairs (ie a sudden drop) does to a vacuum robot so that should not present an issue. Likewise actual steps between lawn areas. Given that the technology has already been developed for vacuum cleaners and is now relatively cheaply available, I would have thought combining the 2 would not add much in the way of cost.

But maybe an alternative system to both would be the answer and one which might be better suited to a garden... I had thought for years that a good system for allowing a robotic mower (or anything else, for that matter) to accurately navigate a lawn would be to have 2 or more (radio?) transponders suitably located to allow the robot to 'ping' them regularly and thus know its location by triangulation. You would need to deal with the issue of power supply to the transponders but they would probably not use much these days and could probably go a year on a small set of batteries.

Then, once the system was set up, the owner would manually drive the mower around the perimeter of the lawn (to 'program' the 'envelope') and then it could be left to its own devices.

Anyway, looks like it's Shank's pony for the time being.

[glnc222]

There was some Asian vacuum robot with a feature to drive around manually to mark boundaries, maybe just a prototype demo, maybe LG's or Ecovacs, not sure, but depending on the lidar or camera guidance mapping. Still on the wish list it seems.

I am not sure radio direction finding is easy to implement compared to line of sight light beams, though I'm not an engineer. Sufficient accuracy could be difficult. Rotating antennas? -- mechanical wear already found on lidar vacuums. It reminds of the Loran and GPS geographic navigation systems. Could be expensive even if possible (though lawn bots are major investments to begin with, similar to riding mowers).
Also this reminds me of the Braava swiffer robot nav with some broadcaster, Northstar, mounted on the ceiling.

It might be possible to add a lidar or something, as a user mod, to detect an existing fence and interface it to the guide wire sensing system in the available mower bots. Sparkfun sells a range finder, without the rotating part of the lidar, $180. Those short range vacuum lidars work by parallax range finding, using a camera sensor to measure angles (instead of light speed timing in radar). Bright sunlight is known to interfere with those IR devices, so probably not good outdoors. Maybe an ultrasound proximity sensor would work -- found on some vacuum bots now, also available in hobby parts.

[edit] To interface an add on sensor to the existing guide wire system, hacking the guide wire signal could be sufficient. A similar signal could be generated in the additional sensor assembly for reception by the robot without touching its electronics. A small faux signal wire near the guide wire sensor. A multimeter frequency and duty cycle scales can read the guide wire signal for reproduction with a simple 555 oscillator IC.
Depends on how the robot uses the guide wires. Monitoring drive wheel rotation gives robots some control over travel.
[edit] The Eufy vacuum cleaner 20 even has an ultrasound, "sonic" boundary marker unit instead of the usual IR virtual walls.
Like an old movie scene "Warning! You have reached the periphery barrier...of his majesty's tulip bed..."

[martinwinlow]

It would, as you say, be pretty impractical to 'hack' an existing machine - unless you *really* had nothing better to do with your life (and if you had that much time on your hands, you wouldn't need a robotic mower!).

Given the advanced state of position-finding tech (thinking GPS, here) the cost and complexity of a radio-based system would be pretty easy (and therefore cheap) to implement. The thing would still have to be able to 'see' (pets, foolish children, escaped ga-ga granny etc) but ultrasonics would do for that. Such a system might even work better than a basic LIDAR system for vacuums, too. The transponders could be simply pugged into mains wall sockets...

It would be interesting to hear form an engineer involved in GPS for their views. The first issue would be what frequency would be needed for the relatively small distances involved. It might be impractically high which would put the kibosh on a radio-based system from the start.

Anyway, back to pushing my machine about - at least it gives me some exercise!

[glnc222]

While not reported yet on lawn mowers, much more expensive than vacuum robots, hacking or modifying the robots and their software is a hobby and a major subject of this website. Engineers, retired engineers and technicians, students all engage in it. Many YouTube videos as well. There is interest in learning robotic technology, similar to the early years of the first automobiles and radio (where amateurs played a major role, "Ham Radio" etc.). The robots - as you have pointed out -- are not perfect as supplied, and additions sometimes make them work more completely in various situations presenting difficulties.

We have not reached the state where debating morality with the household robot a la Isaac Asimov stories supersedes tinkering with the hardware.

I do not think GPS is accurate enough, where inches matter, but I suspect adding ultrasound sensors to a lawn bot is practical and inexpensive as these things go, when already equipped for electronics work with tools etc. I thought these lawn bots had some kind of guidance for systematic navigation pattern, but not so much obstacle detection. I do not own one.
See the "Other Robots" forum here for some threads on lawn mowers.

[glnc222]

Robomow may support software development, not sure what but files at ftp://sw.robomow.com:2120/

[third_deg]

One of the patents I'm part of for iRobot shows the non-perimeter wire system we developed for a lawn mowing robot.

We were to sense physical boundaries by basically making a "not grass" sensor (think "not hot dog" on Silicon Valley).

In short if it sees a sidewalk or pavement or stone walk way, that's a boundary. Same for bumper being used for actual physical boundaries like walls and curbs.

We suplimented with an RF linked chain system for less definable areas. Again, all listed in the patents.

[a1robotrepair]

The GPS model with servo drive control on a zero radius mower shows a lot of promise. These units are currently being used on large farm equipment. More Power, OH oh oh oh OH. Like anything in the 1 off world, very pricey.

[gwleckythompson]

Hello everyone!

It's been an interesting read. Based on my research for my blog, here's what I can tell you about alternatives to laying a perimeter wire to delimit mowing area.

1/ Sensor based robot lawn mowers.

I see that @third_deg cites iRobot patent filings for robots that can sense the difference between grass and 'other' materials, but these have been around for a while. See, for example, the Ambrogio range (from Zucchetti Centro Sistemi). I've witnessed similar in action, and have to say that they do work, but I'd worry that if you had a pond or something similar that you'd be taking quite an expensive risk.

2/ GPS based lawn mowers.

There are a few of these around. Again, Ambrogio have some in their range, as do Robomow and Husqvarna. However, they're not usually used for guidance but tracking. On the other hand, as some have pointed out, you could probably create a homebrew solution out of one of them. Trouble is, they're too expensive for most of us to risk bricking.

Where GPS is a good deal, though, is in being able to train your lawnmower. I remember reading about a make (I forget the details) of mower where you had to guide it around the perimeter with a handheld remote control so that it knew where the boundary was, and, once trained, could be trusted to mow inside it.

I have to say, I probably wouldn't trust the accuracy...

3/ Beacon based robot lawn mowers.

This is where the iRobot Terra (this is not a confirmed name, just something that was filed as part of the 2016 patent application, see: http://www.roboticstrends.com/article/is_irobots_robot_lawn_mower_named_terra) would come into its own.

Rather like an outdoor version of the Roomba system for vacuuming, it would provide a localised, higher accuracy version of GPS, with the only drawback being the possibility of interference.

My personal opinion is that a combination of these, coupled with the usual 'bump' sensors probably equates to a reasonable solution for non-perimeter wire based navigation. It helps to have a fail-safe, and the accuracy of GPS on its own isn't necessarily going to be enough to avoid flowerbeds!

Having said that, the high end mowers used to maintain (for example) golf courses, rely on GPS as laying wires would be prohibitive on such a large scale. Some examples: Turflynx or the Cub Cadet RG3. Okay, so these are not domestic lawn mowers, far from it, but they do show that there are solutions out there that have been developed for situations where accuracy is required and a perimeter wire impossible.

Finally, just so I'm not leaving anything out, check out the Ardumower (http://www.ardumower.de/index.php/en/) web site: if there was ever a DIY way to build a robot lawn mower that didn't need a perimeter cable, then an Arduino based solution is probably it!

However, at the end of the day, there's a reason the perimeter wire is so widely used. It is relatively cheap and well understood, and represents the safest - or at least, most cost effective - way to delimit the mowing area with currently available technology.

[TheNerdyCowboy]

There are a innumerous number of easy, safe, and cheap ways to solve this without fences.

It is marketing. Sadly the vice greed is once again crippling the progress of relieving mankind from tedious labours.

Essentially the products getting the most funding are the ones that are the most costly with the most peripheral costs. That and a horde of zombie branding to make it apear as though there is competition helps maintain the monopolous status quo. The pricing keeps landscapers busy. There are a lot of bodies and a lot of money being wasted on clipping grass and there is a certain lack of competition on many markets these days..