IR single-beam warehouse sensor array design

[iXod]

A warehouse has rack storage shelves up to 20 ft. Local fire code requires that the 30 ft. ceiling cannot have storage closer than 5 ft. to allow proper coverage from the overhead fire sprinklers. A system of sensors to alert forklift drivers of intrusion into this TOS (top of storage) region is needed.

The contractor installing the racks is going forward with IR emitters and sensors from Banner:

     http://www.bannerengineering.com/en-US/support/partref/72971

     http://www.bannerengineering.com/en-US/support/partref/70289

These units have a long history of success.

This installation will have 120 pairs of emitters/receivers spaced 4 feet apart mounted 25 feet up on a warehouse wall. The multiple-frequency option (choose alternate frequency for alternate emitters/sensors) helps avoid cross-talk. And the open collector (both “PNP” and “NPN”) outputs allow each sensor to turn on a light (indicating which row’s TOS has ben violated) and simultaneously pull low (or high) a common “alarm” line which will turn on audible alarms (on a delay).

Wiring for the run of sensors will be 3 wires: +/-24 vdc and sense line.

The only trouble I’m finding in this design is aiming and troubleshooting. If a sensor fails (in a manner that pulls the alarm line low but doesn’t turn on a light) how does one go about troubleshooting which one is the “mad dog”? And when aligning the emitters and sensors initially how to be sure a receiver is detecting it’s mate? A simple switch to toggle power to each emitter/sensor is the easy answer but it’s a bit fiddly to flip a switch 25 feet up without climbing up to do so.

And that’s the big issue: I picture myself scrambling up and down ladders and scissor lifts, back and forth from one end of the warehouse to the other tweaking the emitters and receivers. (And being in the heart of earthquake country (N. California) it will probably need aligning regularly.)

The thought of RS485 comes to mind as a solution. Does there exist such hardware to toggle emitter/sensor power, receive a trigger input from the sensor, and to turn on a light? Where does one find such off-the-shelf product?

Other solutions?

Thoughts and observations welcome.

Thanks,
Dave

[max_torque]

Surely you just need to know, within a few seconds, that an object has been placed that violates the "not closer than 5foot to the roof" rule?  ie, the fork lift starts lifting the (too tall)object, it breaks the beam, and the "too tall" alarm goes off.  At that moment, which ever forklift is lifting an object to the "top shelf" stops doing that?

ie, you don't really need to have a "individual location" based detection, just a simple global plane break detection?

[Ian.M]

Retroreflective tape round the walls at the critical height, and a camera in each corner with some motion detection by image recognition might be a better option.   Mask off (in software) everything below the tape or significantly above it and alarm on any intrusions.  .

[Muttley Snickers]

I think that the OP has indicated that the racking mob has already nominated the type of sensor that they are going to run with and that he has to come up with a way to implement it in a practical manner.

max_torque touched on a good point whereby the limit alert could be incorporated on each fork lift this would obviously be determined by the stock or pallet height. The OP did not mention the warehouse length or the range required for the sensors, personally if I was engaged to install a similar system then  we would use Takex or Optex PE beams, you can run them out to 300 meters, select one of four channels, stagger the RX and TX and utilise the sites security alarm system, easy to align also.

Program a dedicated area or system partition, wire the beams to a local expander or data gathering panel with onboard power supply, localised piezo's or strobes, they can be configured for local alert during working hours and utilised for intruder alarm out of hours, eg: area/on = intruder alarm/ warehouse ceiling area/ zone 52/ monitoring station report. area/off = local buzzer/ time date stamp/ manual code reset or timed output. Individual beam sets should have their own zone and can be isolated or bypassed should the electrician need to change some light globes.

Watch out for temperature, it gets bloody hot in a factory or warehouse up at those heights even on a warm day. Otherwise programmable field lasers, very expensive and limited range. As Ian.M mentioned activity or motion detection on the DVR will work but is prone to reflections into the camera and also clouds if the ceiling has perspex or acrylic sheets, not to mention spider webs.

Muttley

[Kalvin]

Retroreflective tape round the walls at the critical height, and a camera in each corner with some motion detection by image recognition might be a better option.   Mask off (in software) everything below the tape or significantly above it and alarm on any intrusions.  .

Raspberry PI's with wifi dongle and camera module fitted into [3D-printed] enclosure comes to my mind. No special network wiring required. Only power supply needed.

Let's split the $$$$

[Ian.M]

No.  Its a commercial warehouse so an off-the-shelf four camera security system with programmable motion detection would be far more appropriate.  A Pi based system would be less reliable and would need someone who knows Linux, image processing and networking to maintain it.  The one-off hardware cost saving just isn't worth it.

[Kalvin]

No.  Its a commercial warehouse so an off-the-shelf four camera security system with programmable motion detection would be far more appropriate.  A Pi based system would be less reliable and would need someone who knows Linux, image processing and networking to maintain it.  The one-off hardware cost saving just isn't worth it.

Damn. I need to cancel my Tesla order.

[BillyD]

Crude but possibly useful: how about a 20 foot stick with a plate on the end. From ground level you could reach up and block an emitter or sensor. That should at least enable you to verify that the indicator light for each row is operational when you block that beam, provided the indicator is visible from where you are standing.

[SL4P]

it doesn't sound like much of a problem...
Two strong contenders... off-the-shelf CCTV system with adjustable motion sensing windows, or a grid of PE beams, looking for beam break as an alarm condition.

The CCTV will give you a visual record and contact-closure, while the beam break will typically provide a contact-closure.
You can be more imaginative with criss-crossed beams or other strategies with mirrors etc...

But the fundamental solution should be supportable - otherwise you'll spend more on installing, documenting and maintaining it than it's worth. .   if you want to do it yourself, and don't know how - then you may not be the right person for a project like this that impacts safety and compliance issues.

[Muttley Snickers]

Agreed,
It has to work every time and not be a burden on either the client
nor the installation company, do it once properly, no shortcuts.

Muttley

[iXod]

It has to work every time and not be a burden on either the client
nor the installation company, do it once properly, no shortcuts.

Yeah, well that's why I'm asking all you guys. (c;

Video is out of the question. The Banner units are on the way.

No one thinks RS485 or such is a viable solution? Address one sensor at a time to turn on & off for troubleshooting?

Dave

[Muttley Snickers]

I only had a quick look at the link to the Banner IR's when you first posted, What is the maximum distance between walls, I am a fan of RS-485 we use it for ptz camera telemetry but you would still need addressed modules with TTL in and RS-485 out in addition to a front end, that's 120 modules.

Otherwise, and I will have another look at the Banners tomorrow to verify,but if they are TTL out then you can still use trigger relays which can fire on either rising or falling edge, but again you still need a front end, eg; alarm.

Or micro based modules with inputs and ideally lcd display interface and some outputs for alerting device's, battery backed up power supply a must.


Muttley

[Marco]

If a sensor fails (in a manner that pulls the alarm line low but doesn’t turn on a light) how does one go about troubleshooting which one is the “mad dog”?

How about putting a diode on the pull down/up to the sense line and adding a LED visible from the ground which is turned on when the sense line is pulled?

And when aligning the emitters and sensors initially how to be sure a receiver is detecting it’s mate?

The light should go on and off when you wave your hand in front of it :p As for precise alignment to make sure it isn't flaky, a quick google showed Banner has laser tools for it but with a little ingenuity the installer could probably use a laser level to do the same thing.

[SL4P]

As with most alarm solutions - use N/C contact wiring as well... So that an open circuit or connector will appear as an event.

It's surprising how often this is overlooked in designing failsafe solutions. (You can also wire contacts in series with noose of functionality.

[ajb]

Sounds like you'd want a visible indicator of some sort for each detector if nothing else to be able to tell which one has gone out of alignment.  Otherwise, yeah, when the alarm goes off the forklift operator knows to stop lifting and try again.  So you'd want to either wire all of the detectors back to a single panel full of indicators, or mount an indicator at each detector.  I'd think the latter would be best, because it simplifies wiring to a simple daisy chain, gives you a visual indication that you can see from the emitters as you're aligning them, and if you've got multiple forklifts operating at the same time, when the alarm goes off each driver can simply look down the row to see if they're the one who set it off.

On the other hand, having a central annunciator panel would allow you to add a switch to disable each detector.  If the system is really as prone to going out of alignment as you suggest that could be a big plus, so you can mask off a misaligned or malfunctioning detector until it can be serviced while leaving the rest of the system operational.  Of course you could do both a central panel and an indicator at each detector, if you want to maximize the cost and amount of work.

As for aligning the emitters, I don't see any real optical specifications, but given that they're LED and not laser, how sensitive to alignment are they really going to be?  Surely the spot size is going to be pretty big by the time it hits a wall 100' away?

[bookaboo]

Depending on cable distance, impedance, interference etc. you might be able to switch a different value resistor in at each node for diagnostics.
Downside would be it wouldn't work well for multiple faults as you have too many nodes to do a binary type decode.

[SL4P]

On the other hand, having a central annunciator panel would allow you to add a switch to disable each detector.  If the system is really as prone to going out of alignment as you suggest that could be a big plus, so you can mask off a misaligned or malfunctioning detector until it can be serviced while leaving the rest of the system operational. 

I don't want to disrespect the contribution, but disabling safety/fire sensors is not a good idea.
If the solution doesn't do the job reliably, it's probably not the right solution, and may interfere with insurance claims and other issues.

[ajb]

If this were a direct safety solution, then yeah, designing in defeatability is tricky business and requires careful control.  I get the impression, though, that this is more about saving the managers some neck strain from having to watch the tops of the racks all the time, and having a single skid or two above the no-go line is unlikely to turn a mishap into a catastrophe the way that disabling a fire sensor could.

In any case, when the system is going off incessantly because an emitter has failed and you can't get a replacement for another day, would you rather have the whole system inoperable in the meantime, or be able to silence the offending detector and have the rest of the system still work?  Actually, that question applies equally well to fire alarms.  False alarms are a tremendous contributing factor in fire fatalities, because they condition people to ignore alarms and encourage them to defeat alarm systems entirely.

[Marco]

You can just disable the siren until it's fixed, the individual lights will still indicate when the headroom is invaded.

[SL4P]

A warehouse has rack storage shelves up to 20 ft. Local fire code requires that the 30 ft. ceiling cannot have storage closer than 5 ft. to allow proper coverage from the overhead fire sprinklers. A system of sensors to alert forklift drivers of intrusion into this TOS (top of storage) region ?

Sorry, seems that I misread the OP