Optical Boundary Marking Mod

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

Optical Boundary Marking Mod

Postby glnc222 » June 4th, 2013, 4:19 am

TV Remote IR transmitter marks boundaries for Neato with light beam

Roomba employs infrared (IR) light beams to mark boundaries, with its Virtual Walls (trademark, iRobot Corp.).
Neato supplies a different system using passive magnetic strips -- which cannot be accidentally blocked, as one difference.

An optical boundary feature for Neato was tested using a digital IR receiver component, store brand at Radio Shack, similar to many from Vishay and others such as TSOP4838 and other models. 360 degree detection is supplied when pointed vertically -- required when projectors can be on either side of the Neato and both corners must be activated. Data sheet http://www.vishay.com/docs/82489/tsop322.pdf

These detectors receive coded signals from remote controls, indicating which key is pressed on a remote. 38 refers to the carrier wave 38khz frequency, which is pulse modulated at 500hz in specific codes used by audio visual equipment. The tiny 1/4" receiver incorporates digital signal processing and is both very sensitive and immune to interference from ambient light, including strong 60hz components from power line lamps. The receivers are designed for input into digital decoding circuits handling the functions selected on a remote control. They remove the 38khz carrier and deliver the modulated code signal. Resistance to interference includes rejection of improperly formatted codes by the TV standards.

Experiments with plain photodiodes used for IR communications with simple analog circuits could not reproduce anything like the sensitivity and immunity to interference of these integrated circuit receivers, loaded with dozens of transistors and complex circuitry in the way of microelectronics.

A Spot Clean test showed typical Neato behavior for avoidance of cliff barriers.


lightsensing.jpg
Neato with optical boundary receivers, temporary test installation

The detectors are positioned on the Neato top directly over the cliff sensors on the bottom, and are used to activate cliff sensing to input the boundary into Neato's processor. Neato monitors performance of these sensors and predicts effects of maneuvers on them, so detection must be positioned equivalently. Indicator LED's are in the center to show when the transmitter beam is received. These receivers could easily be countersunk into the Neato top. (Unlike Roomba's, Neato's systematic navigation and mapping of boundaries requires separate detection at both front corners, compared to Roomba's single sensor.)

A "virtual cliff" is produced by saturating the cliff sensor IR proximity detectors with IR emitters mounted over the end of the sensor window, or by internal connections to modify the output signal for the cliff sensors (grounding it for cliff detection).
Image

The underside of the bumper on which the IR receivers are mounted is shown below for internal wiring. The ribbed support structure provides space for wiring into the interior. Image

boundary projector

The boundary projector is a TV remote control modified for the purpose. At the standard emitter power levels, 30-100ma through the LED, the receivers are so sensitive reflections off walls deliver the signal, convenient for remote control use but useless for defining boundaries with sharply delimited beams. To define a boundary the power of the emitter must be extremely reduced to around 6ma, in one remote tested, power was lowered by inserting a resistor in line with the IR LED.

[edit] I subsequently learned TV remotes over power emitters up to 500ma (on emitters typically specified 100ma max continuous, less on smaller ones) in short duty cycles to allow cooling, for long range usage. They run at standard frequencies like 38khz. The high power can reflect off walls. Samsung VW's using TV remote tech specify an odd ~minimum~ 2m reach probably for this effect. iRobot systems work at low sensitivity maybe to handle this.

The cheapest 6$ remote at TigerDirect.com takes a 470 ohm resistor. http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=7530743&Sku=O42-1000
The key pad is a single rubber molding held in the case, not glued, easily removed. Power is 3v from two AAA cells.

It is also necessary to collimate the emitter beam into a "ribbon" pattern similar to line-pattern lasers such as picture-hanging levels, to create a sharp beam edge with detection similar to an actual cliff. A reading bar magnifier, 2x, from Barnes and Noble etc., has a 1" focal length and collimates horizontally when positioned this distance in front of the LED (using a small section cut out for complete projectors). The uncollimated vertical spread projects the ribbon pattern on the floor. The emitter is positioned about 4 inches above the top of the Neato.

[edit] a visible LED at high power can be mounted close above/below the IR emitter to possibly get a visible display on the floor of the beam, using a momentary switch.
boundaryprojector.jpg
boundary projector in brown with crystal accents; stand by Kimberly-Clark

LEDcollimated.jpg
White LED shows collimated beam in dim light

The test projector at 6ma produced detection out 10 ft. The power level could be analyzed further to see if longer ranges are possible without creating interfering reflections of a stronger beam. Exact measurement of the power was not possible with the 10hz auto-repeat interuptions of the remote, without an oscilloscope. Results were obtained by trial and error.

The TV remote operates in an auto-repeat mode when a key is held down, repeating transmissions ten times per second. It appears these remotes interrupt the carrier beam in each code transmission, though again no scope was available. Roomba projectors also activate these digital receivers, but appear to send continuous carriers. The receivers appear to reject signals interupted by moving in and out of the transmitted beam, which results in improperly intialized code sequences.

remaining problem
One problem remains with the remotes, in that they shut off auto-repeat after thirty seconds, and must be restarted. Interrupting the power supply results in a less than optimal very slow pair of repeats before engaging the full auto-repeat. Interrupting the pressed key results in immediate resumption of auto-repeat. An additional circuit not shown, with a 555 timer/oscillator component can be used to interrupt the keyboard input to the remote circuitry at 10-30 second intervals. At this time I have still not found a way to connect an external input into the keyboard for this purpose. Any advice on this problem would be most appreciated.

The remote keyboards are membrane switches using a conductive ink on top of the circuit board, connected by some conductive coating on rubber key pads. For a multi-key control board there may be no constant signal to input to a key switch, as both sides of the switch may be used by the logic circuits. A relay can still be used to duplicate the effect of a key press if necessary. It does not appear possible to solder or connect wires to the conductive ink traces, is the problem. It is also difficult to solder taps onto the fine traces of these printed circuit boards from their size. I'd like to preserve this particular remote for possible TV use...

Another remote of interest is a very compact version well sized suitable for including in a final completed projector,
the key-chain remote http://www.ebay.com/itm/Mini-Television-Remote-Keychain-TV-Universal-Prank-Programmable-Replacement-/400484864480?pt=US_Remote_Controls&hash=item5d3ec211e0
and at Amazon. Discontinued by SparkFun.com because of limitations for robotics generally, or supplies. These have a 3v coin battery, best not used. They have a lightly glued metal keyboard cover with snap dome keys easily removed.

projector's own boundary

iRobot projectors include a second top emitter with an omnidirectional lens, running at low power and producing detection at a boundary around the projector, to avoid getting knocked over by the vacuum. A similar feature is needed for a Neato projector, but can be supplied with a different configuration. I suggest a pole attached to a base, with a swiveling arm out a few inches to hold a small transmitter, also swiveling at its end, allowing varied positioning. The transmitter is about four inches above the top of the Neato. The projector base is then positioned behind the beam projection, within the protected region. The area directly below the small transmitter remains uncovered by the main beam, but can then be covered by another emitter in the transmitter bottom pointing down, at very low power. Emitters vary in the spread angle of the light, and the geometry should be observed. The 50 degree spread of the 3mm TSAL4300 may be suitable.
schematicProjector.jpg
Suggested Boundary Projector Mounting


The post can be decorative cafe rod tube topped with your choice of finish and finiel.

[edit] nothing prevents two emitters on the post to mark two sides of a rectangle or triangle meeting at the projector, or in the middle of a 20ft+ line. Power is so low they can be driven by the same source.

At the very low power used, less than 10 per cent of what was found on a Roomba device, small batteries will last a long time in a Neato projector. It is also suitable to power the projector off a wall socket with 3v A/C adapter and left on permanently. Batteries might still require complications such as adding a cut-off time, easily made but more work.

More background information was posted in thread "Adapting Roomba Virtual Walls" (trademark, iRobot Corp) http://www.robotreviews.com/chat/viewtopic.php?f=20&t=16959
As mentioned above VW's do not seem adaptable to Neato's due to the different handling of the carrier signal affecting use of digital receivers. Roomba has it's own signal processing, and different navigational requirements.

other Neato mods
Links to other Neato mods for utility cleaning are listed at
http://www.robotreviews.com/chat/viewtopic.php?f=20&t=16475

[edit] external installation
3v power can be obtained from the vacuum power connector with a $2 buck voltage converter (high frequency switcher) and the sensor electronics installed in the vacuum compartment. Instead of splicing existing wires, try inserting solid wire pins into the top of the plug. http://www.ebay.com/itm/360564973071?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649
Image


[edit] different factory opportunities
A factory system could use the Lidar sensor with no hardware changes to the Neato. Note how the lidar presently detects a pattern on the charger face. Depending on its abilities, the lidar might sense weak coded signals of a particular design from a suitable emitter. Additional software is required for response to detection at the turret in the rear. Maneuvers for obstacles at the front are different, why overhead obstacle detection at the turret cannot be used without additional software.

[edit] part of projector not needed
The extended arm keeping the projector base behind the beam seems superfluous given the local downward marking. I originally thought of it without the downward beam, using the arm as a substitute, which proved inadequate.

[edit]low power measured
DC current through the TV remote emitter at 3v with 470 ohm reducing resistor is directly readable at only 4ma. With 1/2v drop over semiconductor switching, and RMS averaging over oscillation, the average current level operating is likely not more than 3ma and maybe as low as 2ma, for the 10ft range observed.
This DC measure could be applied to modifying other remotes, as a starting point.
Last edited by glnc222 on October 2nd, 2017, 11:20 pm, edited 2 times in total.
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby glnc222 » June 20th, 2013, 12:49 am

Key Chain TV Remote for Boundary Projector

Appears continuous uninterrupted repeats with no time-out, after a minute, so no additional circuitry needed compared to the standard size universal remotes.
20 guage circuit board holes connect with standard solderless breadboard 22 guage jumpers, 2/10" (just inserted; conductive ink, no soldering). Very conductive, easily connects. Dual emitters. Fits in collimating lens housing for projector or base, conveniently small. Lower emitter current as needed for boundary use inserting a resistor in the emitter power lines (see main post).

keychainremote.jpg

keychainremoteinstr.jpg
key chain remote instructions on package


Sold as "key chain remote", "micro spy remote", others, ebay, amazon:
http://www.amazon.com/gp/product/B0054O ... reviews-20
vendor HDE is a high volume dealer in connectors and cables, I've found reliable.

Battery compartment: press short vertical slit sideways in to release tray, pull out.

EDIT note: Photo of collimated beam added to main post above.
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby glnc222 » June 23rd, 2013, 5:27 am

Internal Interface to Neato system

The Cliff Sensor signal cannot be modified by a simple resistor network. Any back voltage on the sensor line damages the device, and the voltages are too low for logic level low's. So mixing with an Op Amp is required.
Analog Distance Measuring Proximity Sensor Sharp Microelectronics GPY0A51SK0F DIgi-Key part no. 425-2854-ND (non-stock). Short distance rated for low power consumption.

schematiccliffsensor.jpg


Inquire for revisions before using any circuit design posted.

[edit] multiply by three limits this circuit to cliff sensors with a low maximum output, and I may have a better circuit with a single dual op amp IC, I still have to try -- looking at a Samsung Powerbot use. Divide the cliff sensor input to the op amp before multiplication, reproducing a full range of input at the output. An inverting op amp circuit might be used on the Mag sensor to ground the other op amp input.

Possible simpler interfaces

schematicAltMagSensor.jpg
Last edited by glnc222 on October 17th, 2017, 10:58 pm, edited 4 times in total.
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby glnc222 » June 24th, 2013, 3:09 pm

Roomba emitters
later noticed one receiver TSMP4138 http://www.vishay.com/docs/82619/tsmp4138.pdf mentions no data format decoding, so might not require interuption of the carrier, allowing use with Roomba VW's unlike the more restrictive one from Radio Shack -- and might not be as good at rejecting interference. Still requires reduction in intensity and collimating lens. Have not tested. Too many varieties of these things, and most popular VW's have radio controls in them not usable. An older "Extended Life" D-Cell model is plain but bulkier than necessary for Neato. Does have a timer, though. Matter of taste.

[edit] another cheap emitter, with a receiver: TV HDMI switcher http://www.ebay.com/itm/5-Port-1080P-Video-HDMI-Switch-Switcher-Splitter-for-HDTV-PS3-DVD-with-IR-Remote-/231007021291
Switch probably useful after removing IR parts. Some of these auto-switch on the HDMI power line similar to USB, from connected device On/Off. Handy. Not tested for IR performance.
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby glnc222 » June 26th, 2013, 10:48 pm

Chapstick Cap Boundary Projector

3/8 inch acrylic rod, IR transparent, cleaves smooth with sharp chisel, for cylindrical lens collimator section, miniature reading bar magnifier; no sawing and polishing the flat. Amazon small pieces. Canceling trip to Antwerp for the diamond cutters, even though lovely this time of year.
Lens optics at http://www.physicsinsights.org/simple_optics_spherical_lenses-1.html
Attachments
collimater.jpg
IR emitter and 3/8" rod section collimator
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby glnc222 » June 27th, 2013, 1:57 am

Mag Sensor Interface

For magnetic personality. Either Cliff Sensors or Mag Strip sensors can be interfaced to optical boundaries. Mag sensors have no plug on the unit, just soldered cable leads to the MCU connector. Plugs with the cliff sensor pin spacing sometimes on ebay, cut to 3 pins as needed -- avoids cutting any wires to splice in. Might be some motion difference in response to the two types of boundaries, not clear yet. The cliff response might not try so hard to closely follow the line as mag strips, was an initial impression; needs duplication.
schematicmagsensor.jpg

Image


[edit] multiply by three limits this circuit to cliff sensors with a low maximum output, and I may have a better circuit with a single dual op amp IC, I still have to try -- looking at a Samsung Powerbot use. Divide the cliff sensor input to the op amp before multiplication, reproducing a full range of input at the output. An inverting op amp circuit might be used on the Mag sensor to ground the other op amp input.
Last edited by glnc222 on October 17th, 2017, 11:00 pm, edited 2 times in total.
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby kpnew » March 7th, 2014, 2:23 pm

Very clever, I like it.
I recently used IR emitters for another project: in Amazon or eBay search for "remote candles"; for under $10 you get a transmitter and a number of receivers...
kpnew
 
Posts: 2
Joined: March 7th, 2014, 1:45 pm

Re: Optical Boundary Marking Mod

Postby glnc222 » March 7th, 2014, 2:54 pm

One set listed $5 delivered on ebay
http://www.ebay.com/itm/3pcs-Flameless-Timing-Electronic-Remote-Control-Color-LED-Vanilla-Candles-Gift-/321343227644?pt=Candles&hash=item4ad18c86fc
Much higher on Amazon.
What's the world coming to, fake candles? Living in virtual reality... Votive candles need a certain involvement, though one button could light a thousand. And AI's could have their outlet.
[edit] short range because receivers hidden inside behind translucent wax?
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.

Re: Optical Boundary Marking Mod

Postby glnc222 » March 7th, 2014, 11:33 pm

Test all circuit outputs and compare to measurements on the Neato before connecting to Neato inputs. The system board's are expensive and easily damaged. A small 100ma fast blow fuse on the 5v power line for mistakes is also a backup. Resistances between power sources and Neato inputs might also protect. Neato sensors must not be subjected to any reverse voltage and connected only to high impedance inputs. These circuits are for users with some knowledge of the electronics involved.
glnc222
Robot Master
 
Posts: 4616
Joined: January 23rd, 2012, 9:19 pm
Location: North Carolina, U.S.


Return to Neato Robotics

Who is online

Users browsing this forum: drude1, Snokia, twistdh, vic7767, zebgelatchi and 363 guests