Maker Pro
Maker Pro

Sensors to detect and locate a fire

Hi, I am looking to make an autonomous robot that can detect and extinguish fire. What sensor should I use for the same? I was thinking a Phototransistor but am not sure if that can distinguish fire from for example an LED light source.
 
Standard arrangement is rather a smoke detector to locate the possible presence of fire. Smoke appears long before fire and much easier to deal with.
Fire break glasses are usually found in fixed locations on major, purpose installed water mains.
 
Fires usually flicker and this component could be used to distinguish from a steady heat source. LEDs however are often driven with a high frequency intermittent supply even if they appear steady. It may be possible to distinguish by filtering, low frequency for fire, high frquency for LED.
 

hevans1944

Hop - AC8NS
Hi, I am looking to make an autonomous robot that can detect and extinguish fire. What sensor should I use for the same? I was thinking a Phototransistor but am not sure if that can distinguish fire from for example an LED light source.
Seems to me that you also need to distinguish deliberate open flames, say from a fire place or kitchen stove-top burner, or even from someone lighting a cigarette with a match or lighter, from out-of-control flames that must be put down quickly. Perhaps an image recognition system, working in conjunction with heat sensors, could be used to sort things out and prevent false alarms. Where would your "autonomous robot" be located? In a home with small children and pets or in an office? Maybe stored away in a niche in a wall until it detects "fire activity" whereupon it springs into action, sounding the alarm, and squirting fire suppressant in the direction of the fire.
 
There is no way you can easily do this. First you'd need to detect smoke, but need programming to understand the direction the smoke drifts, and not have a simple way to determine if the fire was extinguished. Since the smoke will still persist, and drift, and the fire may have potentially spread.

Is the robot going to be dragging around a water hose so it has *infinite* water supply or just an extinguisher that it could deplete without putting the fire out? Sometimes it is not just a matter of spraying in a direction but rather moving or removing things to get to the fire, or cutting off the path of the fire so it does not spread while trying to put it out.

You could end up with a robot that damages things not on fire while not putting the fire out, and getting in the way of a human trying to put the fire out, tripping over the robot in a room with low visibility due to smoke, falling down and hitting their head so they are then unconscious.

I suggest that this is a multi-million dollar project to develop the robot that does less harm than good.

You really should develop your electronics sensing experience doing projects other than those that involve loss of property and life. If there is no human attendant then the time proven strategy is a sprinkler system and an alarm/auto-dialer that contacts a security firm and/or fire department.
 
Last edited:

hevans1944

Hop - AC8NS
You really should develop your electronics sensing experience doing projects other than those that involve loss of property and life.
Yeah, your cockamamie idea sounds good at first reading because there was no critical thinking involved. Who wouldn't want a robotic fireman on duty twenty-four hours a day, every day, all year long?

Of course we have to assume this automaton is at least as well-trained as the human being it replaces. And just as agile. And just as strong. Do you have any idea what it takes to become a qualified fire fighter? Do you have any idea of the life or death decisions a fireman must make in the course of fighting a fire? Do you know how long it takes to extinguish an established fire? Will your robot know how to make these decisions and to work cooperatively with other robots and fire fighters at the scene of a large conflagration? Will it know how to take orders from the Fire Chief? Or will it just blindly follow its own programmed orders and get in the way of REAL fire fighters?

There is a lot more to consider, besides a sensor or sensor system that can reliably detect fires, before even thinking about how to implement and test a robotic fire fighter. And if your "toy" ever makes it to market, be prepared for the law suits that will follow when it detects but fails to extinguish a fire. There will be a huge liability if human life is involved, and it is probably impossible to insure against because of the lack of a track record of success versus failure. So, even if a successful prototype is developed, it may be impossible to market it. But good luck with your idea. I am sure somewhere down the road an acceptable solution will appear, just maybe not in your lifetime.
 
^ Not sure if you meant to quote me or are just intoxicated.

Wanting a robotic fireman and being able to achieve it are two different things separated by a gulf of massive engineering, testing, and liability, to suppose it is sufficient for the purpose which it is probably not.

It's not a realistic goal except for someone who has already mastered the robotics portion and just needs some sensor help and many hours of coding to make the logic work, and even then, can't be comprehensive which is what fire management must be.

A robot that can put out a cherry-picked fire scenario, doesn't cut it. If fires were that predictable, they wouldn't happen.
 
Last edited:

hevans1944

Hop - AC8NS
^ Not sure if you meant to quote me or are just intoxicated.
Excuse me? Quoting you was definitely intentional, no intoxication involved.

Was there something inappropriate about your statement that I quoted? Why would you say, "or are just intoxicated?" I thought I was making it clear that the original poster didn't have a clue about how to design, much less build, "an autonomous robot that can detect and extinguish fire."
 

hevans1944

Hop - AC8NS
The subject of this thread, autonomous robots, reminded me vividly of the opening scene in the movie I, Robot. The character played by Will Smith was saved from drowning when the robot that saved him "calculated" the probability of a successful rescue was greater for Will Smith's character than it was for a little girl child, also trapped underwater in the vehicle wreckage, and most certainly greater than the probability that both could be successfully rescued. So Smith's character survived, albeit with extensive damage to his right arm as a result of being forcibly pulled from the wreckage by the autonomous robot. The little girl drowned.

Smith's character was a policeman and would have gladly sacrificed his own life to save the little girl, but the robot ignored his pleas to save the child. The company that built the robot re-built a prosthetic arm for Smith's character, indistinguishable from a natural arm except it was stronger and presumable "better" than the "meat" version it replaced. This didn't cut any ice with Smith's character and most of the rest of the movie revolves around the tension and mistrust it produced in Smith's character.

So, in the opening scene, Smith's character spots a robot running through a crowd of people, holding a woman's purse high and in front of the robot. Smith's character immediately jumps to the conclusion that the robot is somehow defective and is behaving as a criminal purse snatcher. It turns out that the robot was not defective at all but was trying its best to deliver some medicine its owner had forgotten to take. As the common plot trope goes, "it's complicated."

Fire fighting, or any other activity involving human beings interacting with other human beings, is extremely complicated. AFAIK, the AI community is nowhere close to building a robot, autonomous or otherwise, that even faintly resembles how a human being would respond to a given set of circumstances, perhaps because no one can predict responses of human beings with one hundred percent certainty. There is just too much uncertainty and randomness involved to hope to yield a "clean" solution with the technology available today. Perhaps down the road a quantum computer might be able to solve the problem... but not yet, not today.
 
I think an infrared sensor is a good choice. It can detect the heat of fire and distinguishi fire with the environment, regardless of darkness and smoke. Many fire fighting and detecting devices are equipped with infrared sensors.
 
^ That is an example of detecting above a certain IR threshold without discrimination of what the source is.

You could set the threshold higher than human body temperature but it can't be assumed that a fire is going to be a higher threshold, for example if the middle of a couch is smouldering. It might need to be set to a lower threshold to detect heated air/smoke, to get closer to the source.

It will likely need additional sensor input like optical recognition or at least smoke density, a fair amount of programming, and lengthy testing.

If it were a simple thing to do reliably, someone would probably already be doing it. I don't dismiss the possibility that someone is doing it, but at what cost in money and man-years of time for development? The robot alone, without the ability to reliably put out fires, is a project in itself. How many robots are fire proof?
 
Last edited:
That poor robot will be break dancing down the street with everybody changing channels on their TV or Skybox!.
“Fire!..” nope... BBC1..
“Fire!..” nope.. Volume down..

Martin
 
NO if it's sensitivity is high(too high) it will just be overloaded with EVERY IR source that there is
Maybe you have some misunderstanding about the sensitivity. High sensitivity means that the detector can clearly distinguish the things even they are at similar temperatures. If you don't want to see things at normal temperatures, you can set the alarm temperature over a specific value. Then the when things under that temperature appear, it will not alarm you.
 
Top