Revised December 2013 -- Discoveries From Extended Use On Demanding Carpet
Neato XV-11 made late 2011, used almost two years, replacing a Roomba 500 series, used on low pile carpet. Software has been updated a couple of times to V3.1 at the present with a few refinements in operation over the years. Roomba has since gone through 600 and 700 series with only modest changes, with a bit larger revision in their 800 series just introduced end of 2013.
The most obvious difference between Neato and Roomba is the basic effect of systematic, guided navigation covering every part of the floor only once. This greatly reduces the time spent cleaning an area, by a factor of 4-5 in one report. Whether this results in less cleaning is another issue, but I did not find that so here, at least, just the opposite. And I found the hours required for Roomba to clean just too much. The Neato operates more like the time I would have to take to do it manually. This allows running it when the space is occupied which was not practical with Roomba moving around at random for long periods.
It does not take a powerful vacuum to clean hard floors which are even cleaned with dust mops, Swiffer etc., even available on robots such as Mint (now Braava under iRobot brand).
On carpet, however, dust has to be beaten out of the rug, with power head vacuums etc., and the Roomba did not pick up nearly as much as the Neato with its scheme for applying more power to the vacuum and brush by doing less travel with systematic laser guided navigation.
Various demonstrations of pickup made in reviews with sprinkled rice and what not do not seem to me to capture the real cleaning requirements. I make actual measurements by weight of dust extracted, with microgram scales. These are hard to reproduce requiring actual dirty carpets instead of experimental sprinklings.
My contrast was not quite as large as shown in a finely made video for this at http://www.youtube.com/watch?v=Myh7_0wptd8 , but is basically the same. A full size vacuum still picks up as much as ten times more dirt measured by weight in my experiments, though it would depend on how often Neato is run. My impression is the basic mechanism of Roomba did not change much since until the latest 800 series just introduced now at the end of 2013. I did try the revised intake fan in the Aerovac dust bin with no improvement, at least on the 500 series. The 500 was more powerful than the 400 shown in the video, but still nowhere near the Neato.
Neato is louder than the Roomba as expected with more cleaning power, but can scarcely be heard when running in an adjacent room even through open doors. Carpet does damp the sound some so louder on hard floors, and especially with the original flat beater brush with load slapping noise on hard floors. The helical Pet brush, half bristles as well as beaters, is quieter on hard floors and pulls more pet hair out of rugs than the flat brush. The flat brush, though, has six beaters compared to four elements of the Pet brush.
Neato lacks a flat spinning side brush found on Roomba, necessitated by the intake inset within the round body, but has a wider front intake similar to a regular vacuum cleaning head. There is still an inch or so border along walls Neato does not reach (and Neato does not scrape along the wall, I think allowing room for a quarter round molding sometimes found on hard floors, below the optical sensor for wall surfaces). A different, more expensive VR-100 version of Neato made for a venerable vacuum cleaner company in Germany, Vorwerk Kobold, does have a side brush, but is distributed only where Vorwerk has dealers, parts of Europe and the UK, and not worldwide by mail order, or in the U.S. (to the frustration of many interested buyers here). Nevertheless, I crafted a custom side brush addition to the Neato for experiments and discovered they do nothing at all for carpets, plainly because the brush just sweeps across the top unlike the main brush digging into the carpet. The side brushes have also been found to wear a lot.
These and other experiments and measurements I've made are documented in the forum for Neato at this Robot Reviews website. Many technical details of Neato are covered there for those wanting more in depth analysis and engineering minutia.
An additional problem I had with the Roomba side brush was it pulling out extension cords tucked against walls and breaking off on the sharp corners of bookcases, demanding an endless stock of replacements. So in my case found Neato's wider and front positioned intake a welcome substitute.
Another difference was Neato being able to clean in areas where Roomba's random bumping proved difficult and required blocking off areas, such as vertical blinds and lots of electrical cords. The Neato is able to go everywhere without as much work with Virtual Walls and Lighthouse signaling units iRobot supplies, and small applications of the magnetic strips for blocking Neato suffice. (Only the strips supplied by Neato can be used, as they are magnetized in a single strong vertical direction unlike common stick-on magnetic materials, which cannot trip field sensor electronics at the required distances.) Another nice aspect of the systematic travel is that once finished with an area, it is available for use while Neato has gone elsewhere. And it will not disturb an area until it has finished with other areas, pleasantly predictable.
One limitation of Neato's laser navigation is a short 15ft range of the lidar. It requires landmarks of some sort it can see over that range, so won't work in large empty spaces without furniture to fix upon. More common in homes can be a long unadorned hallway with nothing to distinguish different points along the walls. Roomba will not have this problem. It can be fixed by applying some dark markings at say, ten foot intervals along the walls -- a medalion of sorts -- which the laser can distinguish to locate the robot.
Neato's electronic circuitry seems more substantial, even at its lower price, with a real computer for interpretting Lidar scan data, log files, and communicating for service with PC's through a USB interface (requiring internal components more like a PC or tablet). All sorts of internal data can be accessed for identifying faulty components needing replacement and making various measurements for analyzing performance. For example, the air flow through the filter and the effects of dirt accumulation can even be measured this way. Unfortunately Neato supplies no diagnostic detail on its LCD display besides crude error messages such as this or that is stuck. It still seems a far cry from what I got out of Roomba.
The Neato mechanism seems to me at least much more resistant to fiber contamination than was the Roomba. One of the reasons I finally upgraded after several years use was discovering with some surprise how it failed because gear mechanisms had become clogged with carpet fiber (if not the more severe pet hair problem, truly a challenge to all these devices). There was just a lot of work involved in maintaining the Roomba mechanism, not to mention cleaning the brushes. That is not what I expected or wanted in a robotic appliance, and I regarded that as flimsy engineering not ready for prime time -- more just a robotics demo than a working appliance. The 700 series of Roomba finally addressed some of this with clever detail of the brush coupling trapping hair inside the brush end; see in the forum http://www.robotreviews.com/chat/viewtopic.php?f=1&t=17492  Neato came much closer to expectations for me. That it is not a highly polished shiny toy running on the floor is also consistent with that; it's a tool, not a decoration (though I did decorate it with a two tone color matching protective cover for the Lidar when not in use -- see the Robot Reviews Forum; but it's more like the stereo equipment decor-wise, not a flashy bauble).
The flat beater Neato brush does not need cleaning at all unless in a pet situation, and I expect easier to handle then as well. In the newest 800 series Roomba switched to non-bristle brush design which may also provide relief. But pet owners want the bristle Pet brush supplied for better results which will take some more cleaning. With replacement vanes it is possible to assemble a non-bristle helical pet brush. Various add-ons and newer Roomba models address some of these mechanism problems, but the Neato seemed to have it already, with a completely different design. Not using counter-rotating brushes with their complications goes a long way itself.
Operating Problems and Fixes
Over the two years a couple of limitations on Neato operation where discovered, through reports if not in use here, and simple fixes found for them requiring a bit of crafting with materials to modify the Neato. (Installing a side brush is a much more extensive modification). See the forum for details.http://www.robotreviews.com/chat/viewtopic.php?f=20&t=17372 
There are four navigation problems:
1. Just like the Roomba, Neato can be trapped by failing to sense pressure on the top, from getting wedged under low furniture. Not everyone has such furniture. A simple wire loop taped to the front extending upward handles this and avoids the need to block off or raise offending furniture pieces (and even when adjusting furniture later, allows safely discovering where needed).
2. Neato's bumper does not extend low enough to engage some items such as designer chairs with tubular legs, Scandinavian bent-wood runner leg chairs, very low lamp bases for some examples -- which can trap Neato's when climbed (sort of a Neato Motel effect...). It is inconvenient or not even practical in some cases (like designer furniture) to raise these items enough to properly engage Neato's bumper. The rounded surface on tubes especially defeats the bumper; pack up the beach chairs. The bumper appears to be set so high because of the particular wheel mounting which distinctively lifts the rear very high when bumping something in reverse. This pushes the front bumper edge down which must then be high enough not to dig into the floor. An easy to make replacement for the bottom bumper edge, from common hardware, and a simple thread added to the wheel mount fixes this.
This bumper problem can cause Neato to be trapped on some floor moldings as well, failing to deflect from ones too high for it to cross properly, similar to the problem furniture. The lower bumper also helps with this situation.
3. Neato's tire traction proves insufficient to cross floor height transitions with moldings or carpets in some cases. Very few people have reported this limitation, but it seems fundamental. The mechanics involved with supporting Neato's weight at different points reduces weight on the wheels in these cases, and when the floor or molding is slick the wheels do not grab. Even a low 1/2 inch metal carpet edge molding can be too slick. It was found this can be improved with a softer type of tire for the Neato wheels, made from inexpensive common material. However, there is no room around the existing wheel to apply such, and either the original tire must be permanently removed, destructively, or a replacement wheel crafted working with carpentry tools on common hardware material.
4. Carpet turns out sometimes to be a slick surface on which Neato wheels lose traction, rendering the Neato inoperative there. At the same time carpet creates more drag -- significantly reducing Neato's run time on that surface, specified in the manual. The solution here is not the wheels but a bit of cardboard added to the front of the bottom lifting the Neato slightly out of the carpet and reducing drag (called a slider mod after first using plastic furniture slider fixtures for it). This simple addition makes Neato work very well on these problem carpets (and leaves hard floor operation unaffected). The slickness of carpets may appear only over time, with different experience depending on the great variety of carpet materials and construction. The exact cause remains unknown. Microscopic examination and fancy lab equipment would be needed to know more.
Neato has made a few additions to the firmware over a couple years improving the cleaning operation and what not, and is now pretty much perfected. A few loose ends remain which present only ocasional difficulties. Niceties like more diagnostic info desired were already mentioned. From time to time Neato fails in odd ways, such as running out of power just when reaching the dock for recharging, but failing to repeat the failure subsequently. Perhaps it does not leave enough margin when needing to return to dock, for variations in the demands of travel. Or the computer just gets disrupted from vibration moving around, there is no way to tell. My impression is Roomba's can also suffer such glitches. There is no telling whether this can be improved or any significant additions will be made in future. See the forum for more details. Lots of software revisions can be just for operating revised manufacture of electronics with newer components etc. making no change in functionalities.
One addition to software Neato still needs involves handling of reflective surfaces like mirrors, glass sliding doors, or polished stone facings. Neato's Lidar sees these surfaces as open spaces since the laser is reflected away from the vacuum. Roomba being blind is not affected by these walls. As of the current version 3.1 Neato does not recognize these walls just from bumping into them, and can endlessly try to enter non-existent spaces. If a mag strip barrier is placed the unit will ignore these spaces. Yet the machine has a very usable input in that the optical wall sensor does see these surfaces well, and ought to be processed same as a mag strip when the lidar cannot see it. As long as other walls allow the lidar to locate the vacuum on the floor, it should be able to process position of the wall sensor detection.
In some rare cases lidar navigation cannot work at all: completely reflective baseboards used in some fancy kitchens, completely blinding the lidar. Yet those areas are most likely wet cleaned instead of vacuumed. But Roomba will work there while Neato cannot.
Similar to Roomba in that expensive batteries are needed almost annually, filters and brush replacements. It was found possible to make cheaper filters, for example, saving the frames from used ones and fitting tissue over strips cut from furnace filters. Many varieties of home made filters are reported in the forum. Another inconvenience had with Roomba was the awkward hinged door on the dust bin, compared to the completely pulled out Neato filter serving as door (which many owners have noted). As with Roomba there are aftermarket suppliers for batteries besides the Neato company.
Unlike Roomba, longer lasting and possibly cheaper Lithium Ion batteries, favored for portable computers and power tools, have not been supplied for Neato except in the limited distribution Vorwerk model in Germany. Battery charging is a function of Neato's internal computer, which has to be programmed differently for Lithium, and supplemented with special electronics. There is some prospect now, adding components to the battery packs can adapt Lithium batteries to Neato's normally NiMh battery software, but only partial success has been had, with limited size batteries, so this remains up to investigation.