VX-21 blocked by area rugs?

News and information about the Neato XV-11 Robotic Vacuum. All discussion and troubleshooting questions go here.

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 16th, 2013, 2:55 am

A Tale of Two Carpets

It was the best of times...
http://www.youtube.com/watch?v=okrfS68rlVA

And the worst of times...
http://www.youtube.com/watch?v=LWXzqZsm53E

There is nothing very easy to see explaining why it works in one case and not another. Surprising it works on one of them. Customers will just have to try it on their particular stuff, and try a carpet slider mod when encountering difficulty. I doubt cases where it is totally sunk like that last video can be fixed. A similar sort of rug here was not helped.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby dzmiller » December 16th, 2013, 12:15 pm

glnc222 wrote:thick rug video


Note the inexpensive larger rug Neato easily gets up onto moves and has no pad. Also, the fringe on that rug will be damaged by the beater bar over time.
dzmiller
Robot Groupie
 
Posts: 64
Joined: November 27th, 2013, 4:29 pm

Re: VX-21 blocked by area rugs?

Postby frenchie » December 16th, 2013, 7:40 pm

One more thing about tall carpets. We have a dense high pile carpet that the Neato sometimes has trouble driving on (no problem once it's on it, it always makes it up the carpet, takes a couple of tries), and the edges of the carpet are getting worn out. I believe what is happening is that the brush bumps against the carpet, the Neato tries to push its way onto the carpet for a few seconds, and slowly but surely the edges of the carpet get eaten up ! Eventually it tries at an angle and makes it on, but the damage is done ! The fabric is wool I believe, so it's easy for the brush to grab a whole bunch of fibers.
A change the Neato designers could implement, would be to make the brush assembly (brush + guard) a moving part that would get pushed up by objects that are tall and raise the brush, kind of like the 880. The brush would come down only once the brush assembly has cleared the obstacle, to avoid damage. With such a design, the Neato would be able to go over slightly raised portions of the floor but any significant transition (carpet, door frame...) would raised the brush assembly above it, avoiding wear on both the brush and the transition.
Not to mention they can use this information to slightly lift up the wheels and make going over the transition smoother (kinda of like a tank right now, brute force...).
frenchie
 
Posts: 34
Joined: June 28th, 2013, 8:35 am
Location: CT

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 16th, 2013, 8:17 pm

If you want to do something with that rug it could be surrounded by 1/4" thick edge molding at Home Depot, stuck to the floor with 3M Command adhesive, similar to what another devotee did with mag strips, only here just a ramp to protect the edge getting atop it, and thin enough not to present a floor problem. Needs wood finishing or paint; whether the plastic type can be found don't know. Inexpensive. Decor problems up to personal taste.
[edit] actually is available prefinished in different colors, e.g. cognac
http://www.homedepot.com/p/Hampton-Bay-0-75x90x-25-in-Kitchen-Scribe-Edge-Molding-in-Cognac-KAMS-COG/100556072
More expensive than the unfinished wood.
Try a small 2ft piece by the foot with spot clean to see how performs. Neato turns to the right when starting spot mode. Lot of other molding shapes available, but mostly too thick. These are for exposed edges of shelves and table, counter tops.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 17th, 2013, 3:17 am

Carpet Edge Guard Products

Forget the edge molding above; products are made specifically for protecting the edges of carpet. Most common in some permanent installations for wood-to-carpet transitions, parts of hard flooring systems etc., but looks like some might be adaptable to frame a rug. Should have known from the metal moldings used here on carpet to vinyl floor transitions. Just something to investigate by interested rug owners.

Example: https://www.google.com/search?q=carpet+edge+guard&tbm=isch&tbo=u&source=univ&sa=X&ei=6fSvUviXJvLTsAS4nYDICQ&ved=0CF8QsAQ&biw=1920&bih=934#facrc=_&imgdii=_&imgrc=WbzG9EXCH-vEYM%3A%3BnU0qEkdze0iKZM%3Bhttp%253A%252F%252Fwww.floorandwallsolutions.co.uk%252Fimages%252Fcategories%252Fcarpet1.jpg%3Bhttp%253A%252F%252Fwww.floorandwallsolutions.co.uk%252Findex.php%253Fmain_page%253Dindex%2526cPath%253D259_280%3B391%3B232

Numerous suppliers with extensive variation in dimensions for every situation: http://www.flexcofloors.com/products/accessories/index.html

Lowe's carries Flexco brand, in various colors. http://www.lowes.com/pd_206242-33674-F186V1P071_?PL=1&productId=3022840

3M makes a Trim-Lok brand. http://www.trimlok.com/prod/Carpet-Trim/Plastic-Profiles/All-Product-Categories_143/CP1043_68.aspx

Johnsonite Carpet Edge Guards

Wonder if might be possible to sew these onto the padding at a few points to secure on area rugs. Might just stay well enough themselves. Need to check the product features in more detail.
Any product attached with self-stick glue attached can be handled by gluing it to masking tape for safer and removeable attachment to finishes.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby third_deg » December 17th, 2013, 9:41 pm

Have you tried simply removing the rear casters?

A rear caster is never a good thing for mobility. As it plants the front on the area rug, the rear castors will also take weight, and the weight to the drive wheels is cut down.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 17th, 2013, 10:34 pm

Removing the casters -- just rollers actually -- just drags the back on the floor (with the same weight shifted there); the casters reduce the contact and associated drag [edit] and floor scratching[edit]. The rollers are also in protrusions of the case molding, not attached complete casters, so you will drag points on the floor. Did you try it yourself?
[edit] the wheels are fully depressed with weight on them normally, distributed between the front, wheels, and rear -- no suspension movement involved unless it tilts over an obstacle.
Over obstacles weight shifts to the two ends, with wheels extended by the spring with a measly 1.5lbs each compared to normal, reducing traction. (I put a scale under all the points to compare.)
It's a very plain construction, simplest possible, low cost. Less than Centauri anti-grav modules to replace the rollers.

Maybe you were addressing design generally instead of this specific product? Design a new platform without rear casters, other features changed accordingly?
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby third_deg » December 18th, 2013, 12:03 am

I have not tried as I do not own a Neato.

I was surprised to see that the back end is square, and has rear castors.

At irobot, we designed a tapered rear so that it does not hit when climbing.

Reduces bin volume, but mobility is most important. If you can't get there, you can't clean it.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 18th, 2013, 12:43 am

You must be confusing Neato with another robot seen.
Neato is D-shaped with flat front intake and rounded back to minimize profile for maneuvering, with small non-twisting rollers on short back legs to produce a level stance; smaller front rollers position the front intake at the correct height off but close to hard floors. A problem with the laser guided Ecovacs announced for next year is that it has a more square back. No room in a Neato to add anything, so not surprising.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby third_deg » December 18th, 2013, 12:56 am

No...

By square back I'm referring to the side view of the robot.... Not looking top down (which is rounded)

If you look at the side of any roomba, you will see the back angles up towards the rear, and noting hanging down aft of the rear wheels.

By contrast, the side view of the Neato shows a back that is square edged (again side view). With rollers aft of the rear wheels.

Therefore Neatos first mistake was not allowing a better departure angle in the rear. They then added the rollers to try to fix the first mistake.

Anyway, my guess is the rollers contact before the rear touches down. It's possible for some carpets, just popping out the rollers, might offer just enough improvement.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 18th, 2013, 1:41 am

Thanks for the clarification. Maybe "flat bottomed" might be more descriptive. Engineers appear deeply into subtleties of these shapes than might appear to non-specialists.
The idea seems to be that if weight is taken fully off the rear and balanced on the wheels instead of rear supports, the front can teeter-totter on the wheels more easily, and more weight is kept on the wheels as well. Neato is rear-heavy because of the lidar unit, so will tilt the back down if not supported? "removing" the rear rollers appears to refer to redesigning the whole support platform, not just removing rollers present on the platform in use; I don't see any mod for the existing model.

There is actually very little weight net on the back rollers, why the slightest rear bump flips up the rear. So maybe it would not be so hard to come up with some different design within the space taken by the present wheel motors and suspension (how they even create those things is unfamiliar to me). I wonder if there is some effect of wheel drive torque requiring stabilizing the stance on front and rear rollers. If you make it easier to flip up the front, maybe the cleaning head won't stay down properly. So many factors, you have to see complete solutions, on the non-existent computer simulator.
(There are many tools these days to maybe produce that more quickly than one might expect. Could be an interesting project cheaper than building a complete custom robot.)

It really seems to me to require -- as once commented way back somewhere -- the bots need another motor or solenoid to shift stance, a "height control" or what not, with low down sensors for obstacles, and a more active maneuver to climb over, lifting the front under cpu and sensor control. Much more elaborate and expensive, but that's the complexity of houses and the very small size of the robot. Kind of the "incredible shrinking man" situation. Maybe can be accomplished with sophisticated passive design, which would be impressive.

[edit] if you move Neao wheels farther back to balance weight on them, that increases the radius to the front on the turning-in-place circle, reducing maneuverability.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby third_deg » December 18th, 2013, 9:05 am

Robotic mobility is often characterized by number of motors needed for a given capability.

Differential steer : 2 motors, 2WD turn in place.
Tank steer: 2 motors 4WD, turn in place.
Synchro drive: 2 motors, 4WD, all wheel steer

Etc....

Adding a motor to raise or lower a Neato or roomba differential steer would be 3 motors, 2WD, and would still only be able to work with the weight if the vehicle, plus additional sensors. So would be complex, but still outperformed by any of the 4WD options, that are simpler.

Exceptions being that skid steer is energy intensive, and synchro drive takes up massive volume.

So generally best trade off is differential drive, CG centered over drive wheels with maximum spring pressure for robot weight (trilobite), or CG just forward of the wheels (roomba).

So for Neato (if given a redesign). If Lidar is in the rear, the the batteries should be forward to compensate, and the suspension pivot placed in front of the wheels, with an improved departure angle, and larger diameter wheels. Preferably with the ability to raise the cleaning head passively.

In short, iRobot, and Electrolux placed a higher emphasis on robot mobility, then designed the other systems around it.

Neato placed the bin and cleaning head, and Lidar, then put mobility around it.

As a result it's mobility is not as good as the others. Too me it the most important thing. If you have to keep helping the robot move around, might as well just vacuum yourself.

When establishing the initial engineering specifications for a vacuuming robot, area rugs is a pretty good goal to have.

Neato can fix it, they just have to want to badly enough. Roomba still has it's own issues too. So still waiting for the perfect robot.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 18th, 2013, 3:49 pm

There is always an issue of how extensive are different barriers in homes. Reports are scattered. And we don't have demonstrations of different robots handling the same obstacle.

Neato depends on an obstacle pushing up the front brush guard under forward pressure to mount an obstacle. Works in most cases. Softening the wheel tread helps with some. It could handle softer things better were angled skis or full width plate extended in front of the brush providing a longer lever and over a larger surface instead of thin ribs. There appears to be room for a short form without lengthening the front, the way the bumper is made, but could lengthen the front if longer preventing pushing the brush as close to walls head on, so I've never bothered to try it. I don't have good examples to test it on. Not that hard to make I expect, the attachment being the hard part. Manufactured would just be a revised brush guard shape in the front.

Also, the issue with the carpet is wear on the edge, not failure to climb. It became clear large rug installations are supposed to have edge protectors for all sorts of stuff rolling on and off, from tea carts to full size vacuums. The small throw rugs seem to be the remaining problem. Some get pushed around entirely instead of being climbed, like flimsy door mats. Customers seem to mostly adapt their premises to use the limited robots.

The first video referenced did not have a travel problem but fibers so long they got caught in the brush, from added comments. More often referred to fringes. Could be a problem for regular vacuums with power brushes.

Adding weight to the back might shift the balance to more easily flip up the front. Haven't tried. Tape steel mending plates to the bottom, pretty simple. Even fit inside bottom behind rear grate. It is not practical to reproduce the particular barriers reported casually without details. Even the dimensions would not suffice as the consistency matters, too. People with the problems would have to experiment with different solutions.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 18th, 2013, 10:53 pm

Neato weight distribution climbing
(tilt mod limiting to axle at case edge -- does not appear to effect, as axle above that in photo)

Front raised 3/4". (springs extend wheels)
Total weight 9lbs
Weight on front edge...... 4lbs 10 oz .. 51 per cent
Weight on wheels.......... 3lbs ......... 33 per cent
Weight on rear............. 1lbs 6 oz .... 15 per cent

Flat normal running (case rests on wheels)
Front edge.................. 1lb 8 oz ..... 17 per cent
Wheels...................... 6lbs 4 oz .... 69 per cent
Rear......................... 1lb 4 oz ..... 14 per cent
weights.jpg

Climbing shifts weight from the wheels onto the front, but not onto the rear.

Horizontal springs attach to the short end of a lever arm with several pounds stretch tension reduced at the axle
(Instead of a weaker vertical compression spring over the arm end).

Running flat springs do not support the body weight on the wheels; the lever arm is stopped at the case supporting twice the weight supported only on the spring. Spring tension could be raised almost double till flat running is just beginning to balance on the spring leaving flat running unchanged. How that would affect climbing tilt could not be measured yet. Guessing it would shift less weight onto the front, raising the wheels slightly. [awaiting delivery of additional springs to test]

Tread improvement raises traction 50 per cent, and the supplied springs reduce wheel weight about half tilted.

The wheel arm stops against the case at the wheel end tip, with a much deeper cavity just behind -- space to install a vertical compression spring instead of or in addition to supplied tension springs (which can also be supplemented).

There may be something to tilt first lifting the front onto the plateau and requiring the wheels to then raise a smaller weight rolling up the incline. Shifting weight onto the wheels could increase the load for rolling up, though increasing traction as well. Motor thrust is at least 5lbs from traction tests, but the maximum is unknown.
Lot of interactions affect the outcome. Why experiments are needed (since Archimedes is no longer available -- but I think he'd say the same thing).

Climbing better will not correct damage to carpet edges presenting difficulties, being too soft etc. When protection edges are added the Neato may not only climb as supplied, but the carpet is protected as well.

[edit]Missed one subtlety above: when flat the weight on the wheel is from the position of the tip of the wheel arm, while when resting on the spring, with the front atop an obstacle, wheel weight is entirely from the rear pivot point, transmitted through the arm. Might affect only the transmission of drive torque into the support, not sure how weights work with moving parts, so many vectors.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 19th, 2013, 4:58 pm

Weight Shifting Analysis
weightdistr3.jpg


Running flat, weight is mostly on the front and the wheel axle, divided between the rear pivot and the railed front arm end. When the front lifts and wheels extend, all center weight shifts back onto the rear pivot, shifting a large portion of balance onto the front, as revealed in the previously posted measurements. There is never much weight on the rear rollers, probably from the pivot point so far back.

Increasing spring tension would still position weight on the rear pivot way behind the batteries. While a front pivot does not look practical, compression springs at the front of the arm end would appear to shift more weight off the front, forming vertical suspension. The supplied case appears modifiable to accomplish this, but only with some drilling etc.

One reason an extension spring is used on the rear pivot is to allow a longer spring for proper proportions of stretched to free length, as springs need to move over only at most half the free length, if even that. Still multiple springs could be used for compression. The case structure is not molded for them but the space is there.

As limited by mod for flip reduction, extension varies 3/4" and fully depressed 3/4" space is below the wheel hub rim. The arm tip is over a raised block to which the arm is railed running flat with recessed wheels. It would have to be drilled out for a compression spring mounting tube. Two springs might be needed to get the right properties. There is a couple inches of space under the arm tip (discounting the block) when recessed, compared to the hub rim.

Not sure what advantage such shifts would have compared to increasing the wheel traction. Perhaps easier lifting over soft obstructions. The problem at transitions is slipping against angled slopes, which a softer wheel grips over a wider area. There is also space for custom wheels to be wider than the original.

One way to experiment with these things is use a simple stick with attached weights and supports as a mechanical model of the weight distributions and effects, short of modifying the case.
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 19th, 2013, 7:38 pm

Effect of Cutaway vs Square Back End
(may have other names). Mentioned in connection with Roomba - Neato differences. Does it matter?
(just because the wheel mounting is pivoted to the rear does not mean weight is supported there -- depends on the springs used)
It appears the cutaway could support fixing the axle in place, but I think Roomba has some sprung suspension.
No room for cutaway in a Neato anyway. Neato uses a sprung wheel suspension to shift height of the cleaning head as well as climb obstacles. As they say in New Jersey, You got a problem with that?
Attachments
squareback.jpg
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby third_deg » December 19th, 2013, 8:01 pm

I do! :)

Now perform a free body diagram. Ensure static forces as well as moments are included.

Perform moment and force summations of a robot traveling forward with the wheel pivot in front, and again on the same robot with the pivot in the rear.

With the pivot in the front, motor torque is ADDED to the force applied to the wheel. With it trailing, motor torque subtracts from the force applied.

Said another way. If you hold the robot back with your hand, the wheel is actually trying to drive under the pivot, effectively jacking up the robot.

By contrast with a rear pivot, the robot actually squats.

You have far more traction moving forward with a front pivot.

On trilobite, the wheels translate vertically (no pivot used). This offers greater forward traction than a rear pivot(Neato) but less than forward pivot(Roomba). The advantage for trilobite, is it has the same traction ability driving forward as well as backwards. For Neato and Roomba, it's better in ine direction over the other.

In addition, a robot that can move only it's cleaning head out if the way, does not have to effect robot balance to climb over an object, and can also account the least amount of parasitic losses from a dragging object (any weight not on the tires is traction lost)

So a Robot that suspends it's cleaning head, has a better departure angle in the rear (for better pose control), has a forward castor, and and weight bias, along with a forward pivot, will have superior mobility.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 19th, 2013, 8:53 pm

Might do torque on a stick model with rubber bands around the wheels attached to the suspension arm.
Expand to two parallel sticks for turning. Easier than all the coding for a simulator. Do they have such things at the companies? Or is it all in your head?
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: VX-21 blocked by area rugs?

Postby third_deg » December 19th, 2013, 9:05 pm

If Neato has a test mode (like Roomba), you can prove this out easily.

Simply measure the traction force the robot can generate with a pull test, for driving forward, and again in reverse.

Neato will pull stronger in reverse.

We do establish simulators both for coverage, as well as mobility to answer your question though.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: VX-21 blocked by area rugs?

Postby glnc222 » December 19th, 2013, 10:07 pm

I show just such traction tests in the wheel tread mod thread, but not going backwards -- trivial with a postal scale shown. Neato doesn't navigate backwards much especially to climb obstacles, an effect of the systematic navigation I expect, and does not have any problem there to fix. It attacks obstacles forward and that's where the traction became an issue to users. The torque and pivot position effects do create an excessively high back flip moving backwards which can trap it under cabinets, hence the tilt limiting mod.

Incidentally, there are no fundamental changes in Neato's design since the XV-11, just minor internal hardware changes, colors, brushes and filters. Software tweaks for maneuvering strategies, an area of potential improvement since systematic and open to different algorithms. So you've seen it all mechanically.

Did you get to examine any of the Asian robots?
glnc222
Robot Master
 
Posts: 4538
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

PreviousNext

Return to Neato Robotics

Who is online

Users browsing this forum: allidad, Harry007x1 and 355 guests