Small, Low-Cost Industrial Sensor for Real-Time Object Distance & Length (0.5-0.6m)

I'm looking for a sensor, industrial use, which needs to be relatively small and very cheap. It needs to give the following information:  

1. Distances between objects in real time, as objects may come and an object may leave. This is the red drawings.

2. Object's length in real time. This is the yellow drawings

3. sensing distance need to be 0.5m-0.6 - i.e. distance that can be measured from the black line, to an object. 
.

Sorry for the childish illustration I have added, but I can not send the original files, as they are classified.

Thus, the system needs to know if there is an object over there. If yes, it should tell its length. If not, it should tell where is an empty space, and how long is it. We need to measure only one dimension, and that is the one that is parallel to the sensor. It is very important that the sensors would give information in real time, because the situation is always changing - it is possible that would be many objects at a time, and it possible that would be none at all.

The sensors would be parallel to the objects, as the objects we mean to detect are like cars, which means that the sensorsensors need to be very long. The length of the sensorsensors (the black line), may very over our facilities, so we need the option to buy different length. A good solution may be very very small sensors, that we can place how many that we want.

We would like you to offer us a sensor that you found fit to the following explanation. We are open to hear any suggestion, even if it is a bit different than what we have asked, but one that servers the same purpose in the most efficient way.

Thank you,

Jon  

[User comment deleted]

Hey Mike, 
Thank you for answering! For your questions: 1.  All of the objects are similar. You can take cars as a perfect example to them, they are almost the same as cars. Though, dimensions or the metal that they are made of, may differ, exactly as cars. 2. Of course that as cars, they are made of different kinds of metalalloy. To conclude, if it can detect any car, it can detect any object we need it to detect. 3. Few seconds of delay would be fine - 1-15 seconds. Though, we would rather it to be as precise as possible. Note: price plays a big role for us, so we would rather a very cheap 15sec delay sensor, on a very expansive 1sec delay sensor. Delay would be define: the time between an object leavesget to the black line, and when the system indicates that it has happened. 4. The objects or part of them pose themselves parallel to the black line, and after unknown time they leave or maybe only part of them leave, just as car park. We don't know how rapidly that would happen. But it is very important that the sensor would be active 247, so it is capable to know what happens all time, with the delay mentioned above. As car park, the speed is low - few kmh. 5. Dimensions are the same as cars, and may vary between one object to another. 6. Temperature constraints: 122F - 23F
What do you mean by tickets? 
Thank you!  

it would also depend on whether the objects are moving at a constant speed (then you can use entry/exit to calculate length.  It also depends on what resolution you want, you could maybe use a line of light detectors that is longer than your longest object, that would make it speed independent.  But your detectors would have to be no further apart than the resolution you require. So for a truck-like object and high resolution you'd need a LOT of sensors    It's possible you could combine the techniques - have 2 sensors to estimate speed (time sensor1 - sensor2) , and then use sensor on - sensor off time to calculate length.A bit more information would help.  

Hey David,
First off all Thank you! 
For more information, you can see my comment to mike, if you need more that that, I would be happy to provide it, just tell me what do you need. 
For your questions: The object do not move on constant speed. Though, they move almost the same speed, each one in few kmh, but we it is definitely not a constant value, and we can not predict it.  About resolution, if it knows to indicate yesno - namely, no object there, yes object there-, each 5cm, it would be enough. Cost would be a big part of our decision. dimensions of the object are similar to those of a car. Thus, any object have different dimensions, but we do not know what will come next to the black line. Quote: "line of light detectors that is longer than your longest object", line of detectors sounds like a great idea! About the longitude, the black line is longer(black line - the area which the sensor covers - we need only one dimension.), than our biggest object. Furthermore, we well have to use modification, because we have few locations. and any of them needs to cover a bigger or a smaller area. So a line of small sensors as you said, would be great, because we'll able to choose how many we may use at each location. "have 2 sensors to estimate speed (time sensor1 - sensor2) , and then use sensor on - sensor off time to calculate length." - sounds like a good idea, but is it costly effective? again, we want to economize, we do not need high level of accuracy. 
Another remark that I find important: the object come and exit from a direction which perpendicular to the black like. just as cars park.Regards,Jon

Jon... OK, resolution 5 cm, that's 20 per metre.  Let's say your biggest car is 4 metres long, that's 80 sensors.  And probably 80 signal conditioners, even if these are only CMOS schmitt triggers, and you're also looking at a either a microcontroller with 80 inputs or some kind of multiplexing arrangement, which might involve separate microcontrollers each looking at up to 25 or so sensors each.  That's a lot of work, and a lot of processing, to look at those 8o sensors the whole time and work out the size / speed / distance between your cars.But if all the cars as you say are coming on the same path, and you can arrange overhead lighting and a couple of sensors on the ground underneath them (or even a sideways beam interruption arrangement), it would greatly simplify your sensors.time sensor 1 to sensor 2 will give you speed.time sensor 1 on to time sensor 1 off will, combined with the speed, give you length.and the time sensor 1 off to time sensor 1 on again, combined with calculated speeds of 2 vehicles (or maybe even only the second one) passing through the sensors, give you the distance between vehicles.  It's fairly simple math.This brings to mind a trick that a friend of mine used to play on cops in South Africa (the cops there were renowned for not having too much up top .  They had speed traps consisting of two cables on the ground (capacitive sensing when a vehicle drove over them).  He would ride up to it on his bicycle and lift his front wheel just as he got to the first wire, and get it back on the tar before he hit the second wire, so the system measured back wheel to front wheel time instead of front-front.  As they were only a couple of metres apart, this made it appear as if he was doing about 100 Km/h.  Quite a few times the cops wanted to book him.....

Hhaha nice story! That's how engineers can be criminals too  
Yes I can arrange sensors on the ground, or in sideways. Sideways would be better actually. Overhead lightning beam would be a big problem. And I don't need the speed of the objects, the only thing I need is to get is a yesno output - if the sensor detects something, or not, as simply as possible. I hope it can be more effective now, because as you said it can be very complicated. Can you recommend on a costly effective sensor that is capable of doing that?
Thank you! 

OK, if sideways would be ok, think about a beam break sensor. As you say easier to organise, and not as likely to get disrupted by dirt on the road / floor/track/You have a light generating source - usually a LED - and a detector - usually a phototransistor.They need to be matched so that the LED generates light at the same optical frequency as the phototransistor detects.You can go as simple as a light bulb, an LDR (Light Dependent Resistor) and a couple of pieces of tube, or do it (better) with the LED / PT combo.  And often you will modulate the LED output at a certain frequency, like 40 KHZ, and detect that frequency so you don't get fooled by changes in ambient light.  You'd often need to drive the led at the right frequency yourself (think 555 timer IC)Vishay make a very extensive line of IR emitters and receivers for this and other purposes (TV remotes etc).   In the above link I found this note on receivers and emitters for light barriers which has a bit about what you need and some more references.  If you don't want to design the light beam / barrier yourself you may be able to find some ready built modules to use - if so let us know and I'll try find someone with more experience on that.

[User comment deleted]

JonDavid has suggested privately that I make a contribution here, so if you like what I say- thank him!needs to be relatively small and very cheap. as they are classified.
It is not often that you have cheap and classified in the same discussion! Of course cheap is relative especially if you are comparing to a B1 bomber.At any rate photoelectric sensors are widely used in industry. You can get ones that are reflective or two part with a transmitter and receiver. The name Red Lion Controls springs to mind. Also Automation Direct carry a line. Omega carry some as well. There are several others- just google photoelectric sensors and you will find them like Banner.There is a product called a light curtain which is similar to what you are attempting, except that most of them are used for safety, so they only identify a break, not where it happened. However there may be some out there if you look. I found this document which claims "Standard automation light grids operate using the switching principle. They detect an object at any point in the measurement field and transfer the information “measurement field free” or measurement field interrupted” to the switching output,Q. Unlike a measuring light grid, it can signal the presence of an object but not indicate its position or size." You might want to look into it (figuratively of course, avoid looking directly into laser beams- but you knew that, right?).I have done some work with a thermal camera and provided there is sufficient temperature differential between the background and the object, the outline is quite clear. You may be able to adapt some of the software for your measurements or maybe some other optical recognition with cameras. Oh wait, you said cheap!

The E-mail was sent  
Waiting for you reply! thanks ! 

Beam break sensor sounds great, it meets my needs perfectly. However, I need one part detector, as I can not use sensors from two sides. Thus, one sensor which includes both transmitter and receiver, would be great.  Do you know sensor like that? 
And, yes it would be great to get some help, it is needed! 

Thank you for calling Aubrey! You really help me! 

Hey Aubrey, 
So I assume you are the one with more experience, so after I write that comment, I will reply to David my gratitude. 
Reflective sensor would make it the best solution. I know light curtains, but as they play as a 2 part detector, I can not use them. As David says, I assume you some modules. What reflective sensor should I use? If I use string of them to detect great space, how can I monitor each one gave each signal? Do I need more devices than just the sensors? 
I assume you were cynical about thermal cameras, because there is no way we can afford it  
Thank you so much! 

Jon
 What reflective sensor should I use?I am afraid my experience is limited to looking at adverts in magazines. I have never used one. Perhaps one of the vendors has tech support-If I use string of them to detect great space, how can I monitor each one gave each signal? Do I need more devices than just the sensors? 
Yes, you will need more than just sensors. Firstly you will need a 24VDC power supply (depending on the sensors) of suitable current. Typically the output from each one will be an open collector- so probably easy to make compatible with most input electronic devices. What you use will depend on your experience. You could use a PC, a PLC (programmable logic controller) or some dedicated hardware like even an Arduino as long as you have sufficient inputs for all the sensors.. You could even do it in hardware at the cost of some versatility. In essence you will need to read all the inputs and count the active inputs between the limits that you detect. Of course life always has hiccups and you should write the software to take into account false signals- for instance the angle of incidence of the light beam may be such as to register an absence of an object when there is one there.

Jon... one sided detection.... this will probably reduce your resolution a bit, but not too much I'd hope.  You can do this a number of ways, as other posters have pointed out.  You can still use light / infrared, on the basis that a reflected signal off a near object will be stronger than one reflected of other objects further away.You can also use ultrasonics (think reverse parking sensor type stuff) or if your objects are metallic, inductive detection (like road loops but you can do it that way or sideways).  All these methods would have varying resolution depending on the distance from the sensor (if that varies, so will your resolution).  In all cases, the more similar your objects and the more standard their distance from your detectors, the better resolution you will get.And you can do Light / infrared detection using a mirror at the other side, you can also use those reflectors that you can mount on cars. But I don't know if you are able to do this. 

Hey David,
Do you think inductive detection will be cheaper? And frankly I can not use the other side at all, so mirrors are not an option. 
And do you know models of one-sided detectors? Or what should I tell vendors in order to make them understand exactly what I want? 

OK! So I will talk to vendors.Can you please tell me exactly what should I tell them, in order them understand that I want those reflective sensors? 
Yes, you will need more than just sensors. Firstly you will need a 24VDC power supply (depending on the sensors) of suitable current. Typically the output from each one will be an open collector- so probably easy to make compatible with most input electronic devices. What you use will depend on your experience. You could use a PC, a PLC (programmable logic controller) or some dedicated hardware like even an Arduino as long as you have sufficient inputs for all the sensors.. You could even do it in hardware at the cost of some versatility. In essence you will need to read all the inputs and count the active inputs between the limits that you detect. Of course life always has hiccups and you should write the software to take into account false signals- for instance the angle of incidence of the light beam may be such as to register an absence of an object when there is one there.

Can I learn by the internet how to establish such a system? How complicated is it? 

JonInductive sensing is usually used for very short range stuff - like detecting gears on a wheel.  But it is equally used as large inductive loops buried in roads to sense traffic - we have one at work and during normal working hours you drive over it and it opens the gate (after hours you have to use an RFID pass).  I see no reason why you cold not use one sideways.  BUT the bigger the loop, the further away it will detect, but the less resolution you will have - a vehicle detector is basically a yes/no thing, and you'd still be looking at having the loop within a couple of feet max of the passing metal object.  You can buy them commercially - google Inductive Loop Vehicle Detector for more on that.  You may be able to make 2 small loops a meter or so apart to make an estimate of speed, but it won't be as accurate as a beam break detector for that.  There's a Wikipedia article on it in the link above, read that as well.  I've never done much of that so can't help you much more I'm afraid...

Can you please tell me exactly what should I tell them,Tell them as much as you can about your application. At the very least tell them what you have told us, but more would be better. If you cannot divulge any more, you may be stuck with buying one or two sensors and trying them out. Normally the devices come with an instruction sheet with the connections. You will need a power source (I mentioned 24VDC before)  a Digital Multimeter, and probably a 10K resistor.Can I learn by the internet how to establish such a system? 
You can probably find bits and pieces on the Internet, but not the project as a whole. As with anything, it helps to know what you are looking for so you can ask the right questions. There are also self teaching examples of programming for pretty much any type of hardware that you want n (Internet and books), but in the end there is no short cut to learning how to program. It is only by getting you fingers dirty that you will gain enough experience..To some it comes naturally and others have to work at it. Only the amount of time varies.How complicated is it?
How soon do you need the solution? Forgive me, but based on the questions you are asking, it seems to me that you have some way to go. You may want to get a consultant to help you, but you are going to have to take him/her into your confidence. From what you have told us the project seems fairly simple, subject to the proviso that the sensors actually detect the objects. Sorry I can't be of more help.

OK I will check it out! 
Thank you so much for helping me! You helped a lot! 

I need the solution in the near future, but if a consultantexpert will help as you said, I think I should take one. Because as I understand from you, for me it will take time to *only learn it*, while for an expert it will take few days to solve it - you said it is simple. 
As for the vendors, I have e-mailed some, so thank you for helping me over here! 

In the broadest definition, a sensor is a device, module, or subsystem whose purpose is to detect events or changes in its environment and send the information to other electronics, frequently a computer Processors. A sensor is always used with other electronics, whether as simple as a light or as complex as a computer. In order to explore more about this and to find variety of products-Sensors