Bumper mod question

[rstark18]

I have a few chairs that are just the right size to trap my xv11 (it wedges itself and can't get out).
Can I add a few sticky raised rubber dots (drawer bumpers) to the top of the bumper or will that interfere with the LIDAR? Would black be better than clear? I aw in another thread that someone added these to the bottom of the bumper which gave me the idea for the top.

[vic7767]

Try out a few of the dots and see how the Lidar reacts. Stay with clear for the first test, and let us know.

[glnc222]

Youtube video: http://youtu.be/Vq3RXH6kzxc

(from thread "overhead pinch trap")

Nothing very high can be placed on the top without laser interference (symptoms will be eratic motion). Things very low will not likely prevent overhead traps. But reports of any success will be much appreciated.

[rstark18]

In my case the "trap" only affects the front bumper area so a small riser should do it.
I was thinking black just because it has a hard time seeing black obstacles. I'll try both.

[glnc222]

Beware obstructing laser vision of distant objects, even when the obstruction itself is invisible. The angle subtended by a small thing on the vacuum covers a much larger area several feet away, blinding it to that. But see what happens. If you're underneath the beam and the viewing telescope field, may not matter. Haven't measured it. Might be able to see with any digital camera, they show IR light, looking over the addition at the turret (try it on the wall sensor, for example, positioned just right can see the glowing LED). The laser is only on when cleaning, though -- unless activated over USB.

[glnc222]

GetLDSscan laser readout shows no obstruction up to 5/32" or 4mm above the top surface. Above that the scan shows drops, placing wood strip this thickness on the front. You have about 1/8" is all.

Can't really determine with a camera looking into the laser, you can see from many angles. The beam is too weak for my camera to see the beam spot on paper across the front, but some can -- youtube video showing the line around room walls.

Pushing Neato against furniture shows needed additions.

[rstark18]

GetLDSscan laser readout shows no obstruction up to 5/32" or 4mm above the top surface. Above that the scan shows drops, placing wood strip this thickness on the front. You have about 1/8" is all.

Can't really determine with a camera looking into the laser, you can see from many angles. The beam is too weak for my camera to see the beam spot on paper across the front, but some can -- youtube video showing the line around room walls.

Pushing Neato against furniture shows needed additions.

Good info thanks. 4mm might be enough. I'll have to tak a close look.

[glnc222]

A simpler version of the original overhead trap mod is possible: a square wire loop of say, coat hanger wire -- taped onto the front if not able to drill holes through the screw fastening interior tubes, warranty concerns etc.

The broad overhead wood in the video is for traction against overhead cloth surfaces slowly squeezing like the giant squid recently photographed in the Pacific.
Position the wire cross piece level with the turret top, above the laser beam. Sides must not reflect back into the laser -- requires black tape, heat shrink tube etc. with coat hanger wire, else unsteady motion can be observed.
(thin supports don't affect laser, a long as invisible themselves)

[rstark18]

Well I've been running with two rubber dots on top of my bumper for a few weeks and have not had any problems. It has solved my pinch trap problem.

[glnc222]

Be nice to see what sort of furniture causes the problem in the first place.
With such a small adjustment, looks like raising 1/4" would have sufficed, but is far less convenient -- unless you want to clean underneath, if it would even fit.

[rstark18]

Be nice to see what sort of furniture causes the problem in the first place.
With such a small adjustment, looks like raising 1/4" would have sufficed, but is far less convenient -- unless you want to clean underneath, if it would even fit.

Raising the chairs would have just created more of a trap.
This is the chair that was causing the problem.

[glnc222]

Much appreciate photos. One thing emerging is adjusting the vacuum can be a good alternative to futzing with furniture. The available adjustments have not been very clear in the past.
Same trap problem as other post with wheel damage: http://www.robotreviews.com/chat/viewtopic.php?f=20&t=16382&p=114138#p114138

This example reveals the bumper omits a basic requirement; with a tad more height added for one problem furnishing, now something else a bit higher could cause a trap. As designed the vacuum must be customized to every different environment. Some sort of vertical hit sensing is needed, some up-down play with a switch -- unless this can be mechanically converted into horizontal bumper movement, with some particular bumper shape. Perhaps rollers on the top edge, so if it fits in can't be stuck -- a separate aspect actually. It's a tricky physics problem, and could end up very simple to improve, or require additional sensors.

Needs a little down slope wedge in the front that when pressed trips withdrawal, so it does not depend on any particular height adjustment, but will trip on a variety of heights. This is separate from problems with pressure on the higher turret -- which also has to be detected in the front to allow room to maneuver back.

[glnc222]

Design analysis: if the bumper had a little vertical play supported on fitting with bottom sloping down and back, then downward pressure on the top would push the bumper back tripping the sensor switches, same as a level front push.
Might be room for the fitting below the top surface, where it is presently attached, but haven't examined in detail. When pressed backwards level at the front, ideally it should slide back level, opening a gap underneath over the slope -- requires supporting on a spring somehow. Wide slot bottom for movement side to side, similar to the current wide hole.
A similar effect occurs with mounting on pivoting arms sloped backwards, with the pivot in a horizontal slot; downward pressure rotates it back, horizontal pressure pushes the pivot back. Torsion springs and compression springs.
The initial design was impressive as far as it went, but am still waiting for the follow through. Merely cosmetic changes so far. Be nice if they paid more attention to furniture details. Visit a gallery? Offer one a free cleaning after hours and see what happens.

[glnc222]

electronic fix
The bumper top has a honeycomb structure underneath for strength looks like room to fit thin tact switches at the front sensing vertical pressure, wired in parallel to the existing front switches for horizontal hits. Some rounded plastic over the plunger, with a spring if necessary, not sure, can slide around the cover top supporting the bumper up slightly, and downward pressure on the front will be detected as a bumper hit. Slight variation on the existing bumper switch assemblies, with paddles ("actuators" in the parts catalog).
This is much simpler than any mechanical redesign of the bumper. Detecting a squeeze before pressure traps the unit allows it to withdraw in good order.
Required switches available Radio Shack.

The bumper bottom ribbing also reduces the area of contact sliding over the cover.

underside of top cover, bumper visible through hole

Edit: Ideally the small tact switches (only 1/4" square) should be underneath the top cover under holes filled with compression springs, similar to the existing fittings, with something like nylon screws in the tops as bearing surfaces for the bumper. A portion of the honeycomb bumper underside is removed around the holes for a smooth bearing surface atop the springs, for the horizontal bumper travel. Will have to see if this can be fitted in. Maybe next to the bumper spring attachment post in the photo lower left.

Avoiding turret traps still requires a hoop above the laser but it will now be more sensitive responding to vertical pressure instead of dragging the bumper in.

[glnc222]

Inside Neato's Bumper

Construction of the bumper is shown disassembled. Nylon bearing pads at corners of the cover top surface, and 1/4" deep ribbing of the bumper underside. Vertical sensing switches therefore belong in the bumper underside instead of under the top cover.

Small tact switches from Radio Shack shown fit into the ribbing cavities, and springs from retractable ball point pens fit on the stems to provide more positive bumper support than weak switch internals.
A standard flat washer 1/32" thick fits between the bumper and the nylon pads when the bumper is secured at the top by a lip on the grab handle, showing some clearance. Thin 30 gauge wirewrap wire goes over to the bumper spring pillar holes, and slits can be cut in the ribbing as needed. Limited horizontal play leaves the corner of those holes open for wire passage inside the case, then on to the front bumper switches for connection in parallel.

The wall sensor should be unscrewed when detaching the bumper; the cable connector is very tight and poorly accessible, fragile and could be damaged -- best leave in until the sensor is dismounted.

top cover with bumper removed -- nylon bearing pads

underside of removed bumper -- with tact switch

lift handle which holds bumper top -- tact switch with ballpoint pen spring

The top edge of the bumper is still poorly positioned for vertical sensing as the flange of the grab handle is at the same height and protrudes close to the front, capable of being vertically squeezed, though over little area before a bumper strike would be sensed. To avoid higher turret traps, sensing should really be done at the turret top level, with a hoop on the bumper front. Bumper level strikes would then mostly be horizontal, except at the corners. But Neato does do a lot of turning, working it's way into everything, unpredictable.

Edit: more use for the ball point pen -- the body, sliced section, is a tube around the spring, usable with the low stem switches to hold the spring over the too short stem stub. Then a #6 pan head nylon screw can be used as the top, and adjust the length easier, provide better bearing etc. The pen is mightier than the neato.

Edit: the 9.5mm long stem switches could be mounted under the cover positioned to bear on the bottom of the ribbed underside wells. The cover is about 1/8" thick, the wells 1/4" deep (varies with curve in the cover), and 3/8" is 9.5mm. Only a 1/8" hole for narrow stems, nothing major. Depends on the wells being wide enough for the bumper travel -- not sure. They can be filled in to reach the stem tops. For nylon screw stems only the shaft can be used, for enough travel. Very thin bearing points, but not much pressure on them. Switch travel only 1/64".