slick carpet effect

[glnc222]

Gradual reduction in distance pulling away from the dock when starting, over half a dozen runs, finally required study when failed to move at all -- until turning sideways, banging the dock etc. The motor could be heard with no visible movement first couple seconds.
Turned out the carpet had gotten slick under the wheels at the repeated use spot, and there was no initial traction. Revealed by feeling the wheels hidden underneath, deep in the carpet, while pressing the button -- impossible to photograph. Worked ok after starting out, turning in a different direction getting wheels out of the ruts, like a car in mud. Brushing up the carpet fixed the problem. Likely to happen once the wheels spin on a spot, beginning a polishing operation and laying the plush down in one direction. Each repetition polishes it some more once it starts, with an accelerating descent into the mud. Robotic tow trucks for stuck vacuums?

[JEfromCanada]

This sounds like the issue I faced with my robot, solved with the Magic Sliders. As I recall, you were instrumental in suggesting the solution.

http://www.robotreviews.com/chat/viewtopic.php?f=20&t=16389

[glnc222]

Good to connect all the carpet problems. For deep carpet see link in mods list http://www.robotreviews.com/chat/viewtopic.php?f=20&t=16475
Deep pile could differ from the slick case, fixing Neato on deep and the carpet on low cut. I had reduced traction from slick finish instead of obstruction resisting good traction (which I thought was the deep problem; there is shared sinking in, but maybe different effects -- if the deep was slick, raising the Neato front wouldn't help the way it does?). Tires were slightly tacky as usual. Thankfully not a wheel motor failure.

[JEfromCanada]

I wouldn't call my carpet's pile "deep" at all. The fibres were quite short (about 1/2 inch), but they were so sparse that the robot was "falling through" the pile. Also, because the carpet was slick, there wasn't enough traction under the wheels for it to "power through" the obstruction. Raising the front bumper minimized the front bumper drag and allowed the tracks to power the unit forward, despite the slickness of the carpet.

[glnc222]

Still getting this slipping pulling away from dock on carpet, despite no problem stopping and starting on same material cleaning. I have subjective impression of unusually fast acceleration of the wheels on the pullout, breaking loose the tires, compared to slower acceleration during cleaning stops and starts (it likes to pause and reflect now and then on the mysteries of creation). There could also be a difference in cleaning if start ups done turning, instead of moving straight ahead, complicating analysis. The same area of carpet in front of the dock is traversed while cleaning with no slippage, but there is no stop there to break loose the wheels starting up again.

When the start slips the Neato immediately follows the walls around instead of first meandering out a ways into the room, when it cannot complete the initial spin around in front of the dock. So oddly it turns and runs sideways in the same spot where it has trouble pulling out straight. Doesn't seem to affect the job at all, no confusion. Saves the useless initial meander actually.

I suppose acceleration can be measured by filming the two types of starts and examining frame by frame. Easy for the start up, but a nuisance to catch it mid cleaning. Maybe it is randomized to conceal the pattern algorithm from anyone trying to decipher it from observed behavior. Never know who's watching.

[jin7]

funnily enough im actually having this problem on my Square roboking. On the edge of the rug it loose all traction and eventually gives up and calls for help.
Im not sure its a good idea to add sliders on the bottom of the unit, theres some fairly highish doorsills (~2cm) it needs to climb up on

[JEfromCanada]

funnily enough im actually having this problem on my Square roboking. On the edge of the rug it loose all traction and eventually gives up and calls for help.
Im not sure its a good idea to add sliders on the bottom of the unit, theres some fairly highish doorsills (~2cm) it needs to climb up on

The sliders I used are relatively low, and they also have a curved edge that presents very low resistance to obstacles. I've been completely satisfied with my sliders.

[glnc222]

Little rollers behind the main brush set the front height for the squeegee blade across behind the brush, when on hard floors, but sink into carpet. Sliders correct this to provide the same effect as the rollers for carpet, with a larger support surface, gliding instead of rolling. I can imagine the brush guard should be designed with integral slider material behind the rollers, just as there already is a small center protrusion backward behind the guard.

I wonder if just by publishing such things they can never be incorporated because of NIH syndrome.

[glnc222]

Gliders did correct the failure to pull out from the dock on carpet. Lacking the large round ones, tried the narrow rectangles. Also here the height is positioned with a 3/32" wood spacer slightly thinner than stacked gliders, with the surface about 1/32" below the bottom of the rollers, which then contact on hard surfaces. Foam mounting tape would also serve as spacer.
These sliders look suitable for attachment with the battery compartment screw.

With these sliders the unit climbed a 3/8" metal moulding transition from hard floor to carpet, and a 1/4" wood platform atop the carpet [repeats failed see next post]. It was a bit hard on the metal molding; as found in another thread, Neato can be trapped by a sharp hard edge pressing across the bottom, when it climbs at too steep an angle. There deflecting from the barrier was needed, but here it needs to climb. At up angles the wheels start to lift Neato's weight besides easier rolling along.

Neato's ability to rotate in place depends on using two wheel drive, with tricky consequences for handling surfaces which might better use tracks instead of wheels. The whole thing is not as simple as may appear, just like innards of a fine mechanical watch.
(for other sliders see links in Mods For Utility index: http://www.robotreviews.com/chat/viewtopic.php?f=20&t=16475)

[glnc222]

wrong way gliders
[this conclusion was revised with more data in a following post]

The narrow gliders must be put as long runners to the back; across the front height transitions can trap.
A "V" arrangement pointing into the center also fails (not shown).

correct, runner layout of narrow gliders to climb effectively

Transitions cause a problem also with shorter narrow runners:

half length runners cause traps when climbing

The trap climbing a metal molding is where the wheels engage the moulding, and do not get enough traction.

climbing blockage where wheels meet moulding (prevented with long runners or none)

The long runners lift the wheels up slightly making the climb easier. The short runners may lift the front such a way the upward angle is increased beyond capacity -- hard to determine.

The long runners work well maneuvering on carpet and hard floors. Pulls out at the dock with the slick problem.
The sideways gliders work there but have climbing problems, work only sometimes.

The long gliders still do not improve very sluggish motion on extra fluffy bath rugs; perhaps the round style gliders work better, not tested here.

Caution: do not put adhesive over the bottom label, might remove the printing.

[JEfromCanada]

Caution: do not put adhesive over the bottom label, might remove the printing.

A bit late for mine, I'd say

[glnc222]

Label is coated paper so might be ok. I will put scotch tape across it with a reverse piece under over the label, then stick to the tape surface, fussy me. The bars I would screw on to the battery covers.

When used on furniture I usually nail with countersunk brads as the glue isn't strong enough and doesn't work well on oil finished wood. Gluing stuff to sheetrock very bad as tears off the paper finish, or the glue itself can only be removed with a damaging solvent -- except for the high tech 3M removable glue mounting strips which are stretched to release, great invention. For Neato testing double stick tape is proving remarkably good while removable.

[glnc222]

Another Wrong Slider

This square similar in size to the round ones previously reported fail the 3/8" metal moulding climb test. Also fails when paired on the sides or an additional applied behind deeper, and when doubled thickness.

The only configuration found to climb 3/8" metal molding transition from hard floor to carpet was the long ways bars. However this cannot be consistently reproduced. There is an inherent tradeoff between fixing slick carpet and severe height transitions in these results [found incorrect; see following posts]
They all increase carpet traction; floor transitions is the issue. Not all environments may have these narrow metal moldings.
The minimum spacer thickness for Magic Sliders is 3/32"; 1/16" did not correct slick carpet.

The "magic" brand sliders come with rubber backing and separate mounting tape disks, preferred for screw mounting. Put over two layers of 1" foam mounting tape (scotch tape over bottom for screw attachment).
Their bars just barely fit the screw holes for the battery covers. Others come with thick mounting tape backing, a bit thicker overall, still usable, and cheaper http://www.amazon.com/Waxman-4700595N-Furniture-sliders/dp/B000SL3SEQ/ref=pd_sbs_hg_1 Edit: saw in hardware, look narrower than the Magics, maybe inadequate.
The Magicslider Rectangles 3/4" x 4" http://www.amazon.com/Magic-Sliders-Slider-Rectangles/dp/B000KKXGAQ/ref=sr_1_10?ie=UTF8&qid=1362526755&sr=8-10&keywords=magic+sliders
Might not be found in stores, but haven't looked everywhere.
It is feasible to cut slices of these things.

The square did not improve over rectangles on the extra fluffy bath rug. Doubling thickness of the bars possibly made a slight improvement. Not clear whether anything can be done for that case.

Edit: Dimensions: front rollers 13/32" above the case bottom (another 64th if covered by heat shrink to soften). Magic sliders 9/32 thick. 1/16" foam mounting tape will compress under screw tightening.

3M viscoelastic auto trim double sided tape is thinner than 1/16" regular; two layers gives the 3/32" spacing.
Similar thin tape disks supplied with Magic brand sliders; two layers for the minimal spacing.
Some double sided transparent tape isn't very sticky -- 3M removable; use their permanent type. Tape does not stick well to raw wood; varnish or glossy paint required for adhesion. Glue does not stick well to duck tape, the kind of plastic used. Crazy glue or epoxy only, if even that. Yet there is double sided duck tape, better.

[glnc222]

slider effect on transitions confused

Turns out the Neato cannot consistently negotiate the 3/8" narrow metal molding even without sliders, so the effect of sliders on transitions is less than previously thought. The transition tested is usually blocked by a mag strip in use, and was only used here to test sliders. Insufficient data...
That metal is too slick and narrow, needs a ramp, like edge of a cord protector. But the area is wet cleaned so not needed here. Moldings for wood floor transitions likely wider, wood pieces, forming ramps already.
Some people do Neato the kitchen though.

[JEfromCanada]

By using the round, 60mm Magic-brand sliders, not only do I have a large surface area for distributing weight, but it also presents a very small contact area with the edge of any transition encountered. This allows climbing over any strip from any angle of attack, as long as the transition is short enough to avoid contact with the bumper.

Edit: corrected size of slider

[glnc222]

round slider
As recommended by JEfromCanada who originally tested and explained above, but mounted with 3M heavy duty transparent shipping tape over entire area between battery covers and vacuum cover, with reverse piece under over label (tape comes same width). Two layers of regular foam mounting tape instead of extra thin supplied with Magic Sliders used for proper depth. Thickness needed will depend on the particular carpet qualities.

Modifications for other navigation problems are listed in thread Neato Mods for Utility [url]Mods For Utility index: viewtopic.php?f=20&t=16475[/url]

[glnc222]

Effect On Climbing Moulding

Moulding Neato Climbs (see text)

Neato can climb over the moulding shown at right, 11/16" thick, approximately 2cm, WITHOUT the disk slider installed. WITH the slider, it cannot climb over unless the additional piece at left expands the width of the ramp, lowering the grade. It has no trouble thrusting the front over the moulding; the wheels stop when engaging the edge with insufficient power to lift. The slider lifts the front slightly, resting on the moulding, increasing the angle upwards. This moulding is especially steep. Similar height thresholds will have more of a ramp (seen them in Home Depot 3/4" thick; wall moulding shown above).

[glnc222]

carpet washing effect
After a year+ interval carpet washing (I use a more delicate procedure with Host cleaning material http://www.hostdry.com/, available in vacuum specialty stores, with power applicator), the carpet is no longer too slick for the Neato to pull out of the dock. The carpet had not been a problem for a year. Apparently there is some sort of coating build up on the fibers removable only with cleaning fluid. When dried Host material is being vacuumed up manually, it is unusually hard to move the full size powered brush vacuum head over the surface.
[edit] a day later it's slippery again.

[glnc222]

plastic cutting board sheets

Oddly the slick spot preventing initial roll-out got so even the slider wouldn't help.
A thin 1/32" plastic cutting board sheet, resembling place mats, works sideways taped under around the back of the charger, size fitting well to just past the wheels, not all the way to the front. A strip of hook-side velcro across the back under the charger grips the carpet sufficient to prevent knocking about when nudged by those insistent Neato's following the wall, and preserves portability. Thin material is easy for Neato to move over.

software dubious
There is a software defect in accelerating the wheels unusually fast in the sudden jump outward, as Neato has no problem normally running over this section of carpet. It's doing one of those car play spinning wheel things. Funny compared to inching its way back in to charge. Such an eager beaver. It jumps out and then moves so slowly around the periphery. Wacky.

[glnc222]

Great Slider Results -- With The Right Thickness

Exact thickness turns out crucial to getting the slider effect on carpet. My previous trials may have been a tad too thin. The wheel spinning on carpet proved not limited to around the dock, but slowing the unit down over the entire area, so much it started to take two charges instead of the original one. It could be viewed going at a snail pace especially around the perimeter. Apparently the wheels were spinning loose the whole time.

Raising the 60mm, 2 3/8" slider to 3/8" thickness, resulted in higher running speed all around than maybe originally had before the carpet changed (hard to remember after all this time, and all those mods...). This thickness seems a bit thinner than used by member JEFromCanada -- the first slider user -- who mentioned stacking two disks together. Methods were needed to measure the slider thickness:
The slider top is slightly curved so to get a flat bar across the top supports must be built up on either side, to see when a bar is even with the top center -- or some sort of T-square can be used braced against a vertical side.
Three layers of the mounting tape in rolls 1/16" thick, so three layers are needed (1/32" circular disks supplied with the sliders not used).

A stack of two sliders with two supplied foam disks measured 1/2" thick, quite an increase.

The effect of the slider can be measured placing the unit on a hard surface. A 3/8" slider raises the bottom of the curved brush guard, on which the front normally rests, 1/16" off the surface, measured by a stack of 1/32" cutting board plastic sheets. The edge of the squeegee strip still touches the surface.
When placed on carpet where the wheels sink in even slightly, the brush guard is back onto the top surface of the carpet. Yet the drag is greatly reduced. I see no need to go to 1/2" thickness here, but it could depend on the consistency of the particular carpet and padding.

The brush is not digging into the carpet as much when lifted slightly. There was still dirt extraction but it is hard to measure exactly the effect, without a ton of dirt to weigh. The higher running speed covers a larger area and the unit can be run more frequently. Increased traction also avoids various traps where it was unable to pull out.

The remaining problem may be insufficient glue strength on mounting tape, in which case the epoxy tube must be deployed. Still cover the bottom label with scotch tape shielded underneath -- it may be needed in case of warranty issues.

[edit] velcro for switching between carpet and hard floors