Third_deg rampant speculation thread

Questions and comments about all other Robots go here.

Re: Third_deg rampant speculation thread

Postby glnc222 » February 18th, 2017, 12:26 am

With multiple products serving the same purpose, names need to distinguish attributes of different types. Mobile phones, Cell Phones (shortened to "mobile" etc.) for the distinct mobility feature. So "robotic vacuum", "vacuum robot" needs a catchy abbreviated expression of the same identification. Neato Robotics did well with their "Botvac" (the robot itself another matter). Generic cleaning names can sound like janitor services.

3rd degree sounds academic, PhD, third university degree. Medieval scholasticism -- robotic in its way, intellectually. Or a medical condition.

Apple Computer; the name is like Hoover, purely historical. It had a special connotation of innovation in an era of "mainframe" technology affordable only by large businesses. Rescued by the "i" product names and the historic association with innovation. Fruity robots don't sound very smart AI wise.

There's always CyberDyne... or is that CyberClean? The CyberClean Autovac FloorMaster (not to be confused with FloorMasterPlus follow-on with thigh accessory). How about shoe polishing (unless the market is mainly the Birkenstock crowd -- then the foot massager).
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby third_deg » February 18th, 2017, 4:36 pm

I agree. At the end of the day it's a floor cleaner.

If you have to rescue it all the time (poor automation), then it's no longer cleaning your floor.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: Third_deg rampant speculation thread

Postby glnc222 » February 28th, 2017, 8:31 pm

Battery Issues

Advice from Lenovo on laptop computers says batteries wear out faster when kept fully charged. When laptops are on chargers, a special mode is available not to keep the battery full.

LenovoBattry.JPG


Robots are designed to be kept fully charged so ready for use with maximum capacity at any time. Their batteries then wear out faster than desired. There appears to be a trade off with convenience as well as issues of simplicity in use. More attention could reveal alternatives without all loss of convenience. Battery costs are a concern of customers (Samsung batteries, very large, cost $175).

For cleaning according to a schedule at least, software could be more sophisticated. With the cleaning time known in advance, the battery could be charged automatically only for the scheduled time, and conserved as mentioned by Lenovo. Robots do not even supply a conservation mode to be controlled manually, for fully charging only on demand.

I have not seen this issue discussed respecting cell phones, which are usually not on chargers, more likely with laptops.

(thanks to member shirgal for Lenovo info)
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby third_deg » March 2nd, 2017, 1:20 am

Great feedback! Thank you, we'll look into this
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: Third_deg rampant speculation thread

Postby Tofu » March 8th, 2017, 4:34 pm

Apple's phones and notebooks cycle up and down by a certain amount while plugged in, for this reason.

EDIT: I sort of assumed robots were doing this by now as well?
Tofu
 
Posts: 21
Joined: January 10th, 2017, 6:49 pm

Re: Third_deg rampant speculation thread

Postby glnc222 » March 12th, 2017, 7:04 pm

Unusual intake issues on long pile rugs: http://www.robotreviews.com/chat/viewtopic.php?p=142346#p142346
Dyson climbs the rug and the pile caught in the brush. Something similar reported on Mi without a brush guard. Yet avoiding the rug might also avoid thresholds to climb. Blade brushes? Needs an expert. Can these rugs even be cleaned with a regular vacuum power brush? Back to the old days suction only. On the Dyson canister the brush can be switched off -- but you have to know in advance to do that.
There is still not ability to mark up a map on screen with avoidance areas, and the boundary marker systems remain limited.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby glnc222 » March 12th, 2017, 8:12 pm

Diagnostic software and logging:

Forums are filled with reports of defects or failings with no information available from the complex software to identify possible fixes. Neato robots once had a log file which only the factory could interpret, and then dropped from newer models; other brands offer nothing. When a robot stops on some error, there should be an explanation accessible by usb or WiFi. Very old worn out units are one thing, but these problems appear on newly delivered units or ones under a year warranty. Such is the state of experimental gadget electronics. They have batteries which can be defective and need user replacements, but who knows?. This situation can make a lot of customers loath to even try these devices. Maybe from Best Buy with Geek Squad service, but they don't get any support. If costs are an issue, I guess you are not supposed to buy them (one reason to only get used ones -- maybe they have been proven good by previous owners.)

I suppose it would get boring if everything worked.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby Tofu » March 13th, 2017, 2:14 pm

Are you discussing specifically the Dyson 360 Eye, or are you making a statement about errors and messaging for robot cleaners in general?

I will say that I work in software, and I find it irritating and eye-roll-worthy that when the Dyson experiences a physical issue (caught on a rug, stuck somewhere, thick shag) it gives the maddening message, "A fault was detected". That passive voice is infuriating, not to mention the robot itself knows what's going on, it gives different cryptic light indications for the cause. This is not a company that is steeped in software.

So, my general note to third_deg is to treat error messaging as a key part of the user experience. This is a rookie software mistake to make, but in this rough intersection of software and physical world it will happen a lot, and it needs to be clear, it needs to be informative, and it needs to give guidance on what action the user can take. (There's a ton of material on how to message errors/failures/exceptions/circumstances — Don Norman has some of the best, but all of the stuff out there is pretty straightforward).
Tofu
 
Posts: 21
Joined: January 10th, 2017, 6:49 pm

Re: Third_deg rampant speculation thread

Postby glnc222 » March 13th, 2017, 6:14 pm

I refer to all the robots from reports and personal experience with Neato and Samsung robots. Neato at least has a service mode software system over USB, operating separate motors and sensors, with a graphic interface supplied by a programmer user, Neato Control -- but cannot interpret the error log file, itself discontinued in later models. The robots were good insofar as identifying whether a motor was stalled, for instance, indicating obstructions blocking a brush or wheel etc., but nothing about more mysterious failures, error codes, some of which users have figured out for Neato, lidar failures etc. Yet there are several different codes without explanation by the factory. Only engineering experiments discovered why the Samsung Powerbot gets trapped against furniture legs, with help from experience with the Neato (the wheel extension sensor for lifting the robot off the floor getting tripped when the robot tilted, and a programmed response of shutting down -- at least on the Neato the sensor is a visible microswitch, with data shown by Neato Control service software).

Beyond simple failures is tracking the wear on the battery. The software in Neato at least, appears to adjust the run time of the unit over time as the battery capacity falls from wear, maintaining a reserve to return to base -- yet does not report this information (there is a way to get run time tediously on the LCD display with unpublished interpretation of binary data). The Botvac has a fancy MAXIM hardware chip to measure battery charge, Fuel Gauge, integrating IxT. Yet the only way users discover the battery is worn is all sorts of system failures from inadequate power supply, indistinguishable from hardware defects. I even fiddled with a hardware circuit addition to indicate battery wear, using a clock timer chip, but found it more practical to just mount a kitchen timer on the dock sensing departures and arrivals of the robot at the dock. Threads in the Neato forum, Mods & Repairs index main page.

The kinds of features needed in the software may have to come from the designer familiar with all aspects of the product, to properly manage software experts -- or outside contractors -- who implement the features. It is a management issue, how seriously do they take the whole enterprise.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby glnc222 » March 13th, 2017, 7:06 pm

Back on the cliff sensors: if not aware, there were "keep off" mats and strips sold to exploit this defect for boundary marking with Roomba's, though reviews said only partly effective. One source of material for testing. Most common black color things are still visible in IR. It takes expensive stealth paint sold for this as well, though one type did not work on the bright cliff sensors -- for defeating lidar speed detectors.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby glnc222 » April 16th, 2017, 11:38 pm

Again on the dark floor drop sensor problem:

Just recently observing how the cheap vacuum lidars work by parallax unlike radar timing (other post http://www.robotreviews.com/chat/viewtopic.php?f=3&t=19887), I can imagine a possibly inexpensive lensed parallax detector specialized for the drop limits only, where any reflection detected within an offset, focused receiver is good and anything deeper will not be seen, outside the detection angle -- binary detection only. So no expensive camera sensor like full range finder. Issue is whether an intense enough emitter can get a reflection at all from dark material. Avoids the issue of intensity based precise measurement used in standard proximity detectors, angles instead of intensity. Simplifies whether emitter is laser or led.

The parallax method lidar is made possible by the focused spot from lasers compared to simpler emitters, avoiding pattern recognition expense. Lasers used in optical computer mouse pointers, too. But for short range a lens focused LED might suffice, depending on specs, maybe just not the particular lenses usually made. Maybe lenses added to the sensor cover plate (like I added a collimating lens chipped from a plastic rod to Virtual Walls). Tough the problem carpets and flooring not common to reproduce the difficulty. Lot of dark materials work fine.

[edit] Were a standard proximity detector be adapted to perform angle detection I would wonder if the ones lensed for longer distances might be more useful than the short range ones normally used for drop sensors. The optics have to be worked out.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby third_deg » April 17th, 2017, 6:08 am

Very cool thought. Seems very plausible. The challenge would be to make inexpensive enough relative to the rest of the device.

I'll definitely think more about it however.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: Third_deg rampant speculation thread

Postby glnc222 » April 17th, 2017, 1:02 pm

Maybe some plastic molding cover on a standard sensor part forming prisms or light pipe on the emitter to shine focused from a small offset at an angle, receivers already focused. Neato uses Sharp GP2Y0A51SK0F short range, at Digi-Key.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby Peter101 » April 17th, 2017, 1:34 pm

Neato lidar uses a strip image sensor that can tolerate bad pixels. See the attached paper and link
https://pdfs.semanticscholar.org/4d8f/5 ... 677254.pdf

The Neato lidar protective lid has support pillars which creates a few small blind spots for distance ranging.
A mirror or a light pipe could deflect the optical path down where a cliff sensor would be needed.

-Peter

Neato Lidar main.pdf
(361.54 KiB) Downloaded 27 times
Peter101
 
Posts: 19
Joined: January 18th, 2016, 5:57 pm

Re: Third_deg rampant speculation thread

Postby glnc222 » April 17th, 2017, 3:59 pm

Very creative theoretically, but the lidar is weak for safety, large light pipes could be expensive, and maybe absorbing. I don't think Neato/Xiaomi lidars see dark surfaces. Third_deg also planned camera guidance like the other brands other than Neato and Xiaomi. The camera guided bots though, are festooned with surrounding proximity sensors for obstacles, raising costs. Samsung has some powerful emitters on what looks like plastic light pipes (maybe sharing emitters among multiple receivers, or just mounting convenience), haven't examined fully. Their receivers can track a spot projected on the floor by the remote two feet away (but still bumps dark objects).
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby djos » April 17th, 2017, 6:47 pm

glnc222 wrote:Very creative theoretically, but the lidar is weak for safety, large light pipes could be expensive, and maybe absorbing. I don't think Neato/Xiaomi lidars see dark surfaces. Third_deg also planned camera guidance like the other brands other than Neato and Xiaomi. The camera guided bots though, are festooned with surrounding proximity sensors for obstacles, raising costs. Samsung has some powerful emitters on what looks like plastic light pipes (maybe sharing emitters among multiple receivers, or just mounting convenience), haven't examined fully. Their receivers can track a spot projected on the floor by the remote two feet away (but still bumps dark objects).


From what I've read from owners of the camera based nav bots it's seems they are far more easily confused than the lidar based Neato's, especially in poor light or changing light conditions. A better approach imo is the Vorwerk style which adds ultrasonic sensors to the front bumper to complement the lidar.
Current Robot's:
Neato BotVac 80 w/ Li-NMC Battery , 2x XV11 w/ Li-NMC Battery
iRobot Roomba 561 w/ NiMh Battery & Aerovac Bin, Scooba 380 w/ Li-NMC Battery

Previous Robots:
iRobot Roomba 562PE w/ Aerovac Bin & Sealed-bearing CHM
Evolution Mint 4200
User avatar
djos
Robot Addict
 
Posts: 1202
Joined: November 30th, 2009, 7:34 am
Location: Melbourne, Australia

Re: Third_deg rampant speculation thread

Postby glnc222 » April 18th, 2017, 3:17 am

Further drop sensor simplification:
when using angles instead of intensity, the normal proximity circuit is not needed and sensors might be made as fairly simple little circuit boards, instead of adding optics to standard sensor parts with integrated digital circuits.

To distinguish a reflection from background illumination, only an oscillation of the signal need be detected, just by passing through a capacitor removing the constant background -- since the amplitude which makes up the complexity of the standard circuit, measuring intensity, comparing background with illuminated etc., is not needed, angles being used instead. No local microcontroller, just an oscillator etc.

An emitter LED can be mounted on a wedge of some kind on a board for an angle or inserted into a configuring plastic cap on the board aligning the optics, like holes drilled into plastic sheet. As hobbyist only I have never figured out what components make really sensitive detectors like in the integrated coded TV remote receivers which pick up virtual walls thirty feet away, so need a professional. Darlingtons?

A weird thing: Samsung using TV type sensors for their abandoned VW's required a minimum instead of max distance in front of the WV because reflections off walls would screw it up. They switched to mag strips last year. Neato used little custom boards for mag sensors, so why not drop sensors as well? And you can file another patent entertaining the lawyers.

[edit] On a vertical board LED's can just be mounted on holes at a slant, no fancy plastic parts. Combine with mag sensor same board in space between emitter and receiver. I think I will try making a prototype on perf board, include an indicator led, and take to IKEA in search of special black carpet.
Last edited by glnc222 on April 18th, 2017, 2:37 pm, edited 2 times in total.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Third_deg rampant speculation thread

Postby third_deg » April 18th, 2017, 3:20 am

Agreed (djos),

We will be solving via sensor fusion in another way. Also, the method of movement has other key navigation advantages that other 'bots can not do, which helps in a very large way.

I've been generally impressed with many aspects of the vorwek however.
third_deg
Robot Master
 
Posts: 507
Joined: November 12th, 2013, 2:01 pm

Re: Third_deg rampant speculation thread

Postby djos » April 18th, 2017, 4:27 am

third_deg wrote:Agreed (djos),

We will be solving via sensor fusion in another way. Also, the method of movement has other key navigation advantages that other 'bots can not do, which helps in a very large way.

I've been generally impressed with many aspects of the vorwek however.


Awesome, I'm really looking forward seeing your final design! 8)
Current Robot's:
Neato BotVac 80 w/ Li-NMC Battery , 2x XV11 w/ Li-NMC Battery
iRobot Roomba 561 w/ NiMh Battery & Aerovac Bin, Scooba 380 w/ Li-NMC Battery

Previous Robots:
iRobot Roomba 562PE w/ Aerovac Bin & Sealed-bearing CHM
Evolution Mint 4200
User avatar
djos
Robot Addict
 
Posts: 1202
Joined: November 30th, 2009, 7:34 am
Location: Melbourne, Australia

Re: Third_deg rampant speculation thread

Postby glnc222 » April 22nd, 2017, 3:05 am

Test of a parallax cliff sensor with a phototransistor. http://www.robotreviews.com/chat/viewtopic.php?f=4&t=19897&p=143276#p143276
Black knit cloth was undetectable despite other black materials working. Only a much more sensitive receiver would work, maybe a camera sensor. They are used in the parallax lidars which do have weak returns, but I do not know specs. Yet with high enough sensitivity, emitter reflections between the floor and bottom of the robot could present a problem achieving the necessary discrimination.

It appears optical cliff sensors are destined to be disabled in some dark floor situations. Samsung says as much in their user guides.

While users easily solve the problem by covering the cliff sensors, there is a human factors problem in supplying a standard setup option. If improperly used, there could be product damage from unexpected tumbles. The sensors must be hard to disable even if possible. Pop up a lot of warning and confirmation screens in any menu option for disabling.
glnc222
Robot Master
 
Posts: 4582
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

PreviousNext

Return to All Other Robots

Who is online

Users browsing this forum: aarolar, calebleemcd, Franzl, jramer, leonicholson, longerski, Snokia, teokakav and 484 guests