There's a safety requirement in IEC 60825-1 that under any foreseeable single fault, the light output can't exceed eye-safe intensity limits. It's an issue particularly with IR LEDs and lasers because you don't have the natural aversion to staring at a bright light. The microcontroller duty limiting circuit is very likely intended to meet this. You'll find similar design considerations and current limiting in LED and laser mice.
Not to mention doing it that way also have a nice side effect; derating any of the semiconductor light emitters also ensure that they last much longer too.
I doubt a pure software solution will meet that "functional safety" standard. Thus there should be a design aspect that stops runaway light output even if the software microcontroller runs amok. Maybe the switching transistors have a pulse width limiting driver circuit, or maybe they rely on frying the LEDs or resistors before they fry anyone's eyes.
Heehee! I used to work at Microsense! (I was a programmer working on their comms, monitoring systems, and control, rather than detection, which was quite specialised). We were always a little bit of a minnow around some much bigger sharks, but it meant we often got 'interesting' requests from customers (mostly local authorities) that the bigger companies didn't have the flexibility for. One of our higher ups on the team was heavily into optics, and did some great fun work on both detection and signals, improving over the older systems as LED technology started to take over. There were several AIM detectors in the suite, with the LEDs configured for roadside, gantry, kerbside and stop line use.
@@samverve Via Traffic Signals UK, and Traffic Signals EU (who sequentially took on the MSL brand first). I was there for the end of MSL, all of the above, and tiny a bit of telent, before moving away and up to Glasgow. Over the years we got up to some fascinating stuff in signal, detection and control. And my enduring memory is the internal part number for a waterproof toggle switch cover for the controller manual panels, which was ME-NIPPLE. :)
Fun trivia: The term 'Pelican crossing' comes from PeLiCon - a contraction of 'Pedestrian Light Controlled'. This type of crossing has two fixed time periods: a steady green man / red light to vehicles (indicating pedestrians may cross / vehicles must wait) followed by a flashing green man / flashing amber light to vehicles (indicating pedestrians may continue to cross, but must not start to cross / vehicles may proceed if the crossing is clear). The term 'Puffin crossing' comes from PUFIN - a contraction of 'Pedestrian User-Friendly Intelligent'. This type of crossing does away with the second fixed period (i.e. the flashing green man / amber) and instead uses the sensor, shown in the video, to detect if pedestrians are still on the crossing - if so, it extends the initial steady green man phase until the crossing is clear.
@@douro20 If you're in the UK, puffins are the ones where the signal that it's safe to cross is annoyingly placed in the same unit as the call button, instead of being on the pole on the opposite side of the street.
Clive, the inner 'infrastructure nerd' in me loves things like this - it's a long shot but would love to see what's inside a speed camera. Definitely love to see more street furniture and infrastructure!
Brings back memories. Used to work on traffic lights years ago. There's loads of cool tech in traffic. Doppler radar, video detection, crossing tech to help vision impaired people, emergency vehicle priority transponders.
Good old AIM-K detector. Used to work for Microsense and these were a really good kerbside detector for use at Puffin Crossings. These self learn on power up and you keep the area clear so they learn what a normal clear crossing looks like. This one looks like a later unit manufactured in Hungary by videoton the same people that make televisions and other electrical products
Hi Clive, when you said "People don't stay still at lights" is why I kinda think I know how these things work, usually you can see the red LED that shows detection as a pedestrian (in the ones I've spent time looking at!). If you're a wheelchair user and you don't move, it will cancel when you've pressed the call button, which is quite annoying. I'm pretty sure they use these to stop people running up, hitting the button and stopping traffic.... so annoying kids basically, but yeah, now I have to wiggle when sat waiting at lights or it sometimes places me in the 'background decoration' category :p
I guess the detection is done by lock-in amplification: When the LED is on, the ADC value from the sensor is added to a variable. When it´s off, the value is subtracted. This is done for a couple of strobes. This eliminates external light disturbances and indirectly amplifies the light of the LEDs that an object is reflecting. Changes in the magnitude of the reflection can be easily registered.
Correct, there are 2 types of detectors, On-crossing and Kerbside. As you can probably guess, the Kerbside dets looks for peds at the kerbside that have pressed the wait button, if the det no longer detects peds at the crossing it can cancel the wait button demand (as its called) . The On-crossing detectors look at the crossing itself and will hold the lights on green (for the pedestrians) up to to the max green time if it detects pedestrians still on the crossing eg lots of people crossing, elderly crossig slowly etc.
This is genius, I don't think we have anything like this used in the U.S. It's all just pretimed from what I can gather. Hit the button, it eventually gives you a crossing signal, some have a countdown clock, others will begin to flash the pedestrian symbol about 10 seconds before it turns back. I like the idea that the system can adjust for slow or fast walkers.
Definitely could use one of those auto-cancel features in many ped/bike crossings here. People just push the button, ran through the red anyway, and let 15 cars sitting through a minute at red.
@@greentjmtl there are a number of crossings near me that have sensors to detect if there is still someone waiting, unfortunately, the sensors aren’t perfect, if you are stood too still, they don’t see you, quite a few times ive been stood waiting at one, and it has cancelled, so i end up having to wobble underneath the sensor
I just love these careful, clever analyses by BigClive. The hardware design is brilliant - especially with the special dispersion pattern of the LEDs and IR receivers - but Clive hit the nail on the head several times, especially with the "magic" software, which IMHO is reminiscent of advanced digital video synthesis on earlier radar systems, which wrestle with the same detection and measurement problems in terms of determining what is ground clutter and filtering it out in a smart way without losing the ability of detecting and tracking objects "hiding" in the clutter.
When it comes to hiding the clutter, it somewhat reminds me of those detectors for automatic door controlls, those usually learn as well to workl somehwat predicatably. otherwise even having for example a large dark floor mat on a lighter tile or floor (which doesn´t seem to be a rare instance) could set them of. If you are patient enough and able to stand quite still you can test it, if you remain very still in front of them, at some point they will stop onening and closing and remain closed.
I remember that we had a unit just like this (or highly similar at least) in the late 70s as part of a very high-end home alarm system. The chips might not have been a part of it, might all have been done with discrete parts though...
The ir detectors used to be added to junctions that had issues with cyclists and motorbikes not being able to trigger the induction loops due to the alloys used to make them. By pointing them just in front of the stop line they allowed them to trigger the junction and get a green light.
They've been using video detection for this purpose for about last 25 years here (Virginia, USA). The first system they used was the Econolite Autoscope system. It had the video processor in the same enclosure as the camera. It used an X86 CPU, think it was a 386. There was a broken one at a job I worked at, but I never got a chance to tear it down to see what was inside.
The Infrared emitters in this pedestrian sensor may actually be a VCSEL diode rather than a LED, as I took a look at the emitters where you got the camera close enough to the board, I could deduct that it is a laser diode due to the fact that the VCSEL diode die looks fundamentally different than the LEDs when you look into the LED package, that is with the electrically connected laser optical cavity in the center of the die (and the wire bonds tend to be welded at the corner of VCSEL die). It could be a 940nm Infrared VCSEL, who knows?
they do look odd and being labeled as "LD#" suggests to me they're laser also. also, it seems the camera does not pick them up at all. even when he mentioned a glimmer, it looked like nothing but reflected ambient lights.
@@throttlebottle5906 Human eye can detect pulsed IR light up into the 1300nm (iirc), while the hot mirrors on normal cameras will reject almost all 850nm and almost almost all 950nm already. If these are 1060nm or 1100nm VCSELs, then it'd be probably entirely invisible to most cameras but the eye can still sense it.
Another fun fact about crossings is that they are all named after birds. Pelican, puffin and toucan meaning 2 can cross pedestrians and cycles. Also Pegasus which is for horses but also has wings to fit with the bird theme. A cuckoo was a term used to describe a control system from another manufacturer fitted in the wrong case.
@@foxxy46213 bit of a stretched acronym. pedestrian user friendly intelligent. A puffin puts the red/green indicators above the button and makes you look in the direction of traffic to see them.
The marking around the infrared LEDs uses the abbreviation “LD“ instead of LED. Usually when I see that… It stands for “laser diode“. Which would make a lot more sense because they would be much more focused and are dirt cheap these days. I would think maybe it was originally designed for low power infrared laser diodes but they found they could get away with LEDs. Just a thought. Also the photo diodes are properly labeled and abbreviated “PD“. I’ve seen similar set up an alignment used in those laser tire alignment machines used in automotive shops. I’ve had to repair a few of those for random clients and they use almost the exact same setup just a bit more advanced.
Yep, I agree, it's a laser diode, since I already can tell, having handled both regular LEDs and VCSELs in the same 5mm LED epoxy packaging (they look the same except when you take a look at the die in it through the lens, it's going to be obvious here, they look different).
It is looking for the change in the pulse signals from the emitter. By looking for the pulsed signal it is not sensitive to slowly changing IR from sunlight.
I remember back in the 90s... one particular crossing would give false negatives all the time... I took to jiggling around at that crossing until they got the calibration sussed out.
I think that rather than cancelling the call, they actually come into play once the crossing has changed in favour of the pedestrian. It holds off the change back to traffic mode until all the pedestrians are clear in order to aid less able users. They also allow quicker return to traffic mode if an athlete scurries across. In general, once a pedestrian has pushed the button, the lights will cycle regardless of whether the user waits or not. The scallywags round here used to stick chewing gum in the button and leave the lights to keep changing, showing that there is no call cancelling.
The LEDs may be modulated to discriminate it's own signal from steady state IR sources (like sunlight). This is a very common technique, even consumer IR remote controls do this. I remember at a long-past former employer we had a person tracker device to allow a pan/tilt camera to keep the person centered. The person being monitored (in this case a patient who was also connected to our EEG equipment) would wear IR emitters. The emitters were modulated and this mostly worked well to keep the camera from locking onto the window when the patient moved past it. Until some of the hospitals that our equipment was used in changed their fluorescent lights to new ones that coincidentally had the same (or similar enough) frequency as the modulated IR that we used for patient tracking. Had to do a quick design change to change the modulation frequency that we used in every unit that was affected.
I still remember when I was a kid,standing outside the Royal Albion Hotel in Brighton,in the middle of the beam controlling the traffic lights. I still remember my Dad coming around the corner,too!
Hello from North Carolina in the US, Clive. I'm Eric another Friend of Dorothy, and fan of your interesting explorative and expansive videos -- and now your PINK CALCULATOR also! I will try not to be jealous, kind sir, just as you should avoid falling under the spell of the green-eyed monster when I tell you about my Pink, Crystal Studded HDMI cable, quite possibly the gayest HDMI cable found on any of the inner planets. My background is in software (more specifically, software PROCESS and Design Engineering) so I lack a lot of basic electronics understanding and eduction but I still find your videos very interesting and informative. Please do keep on doing what you do. Your videos are compelling for those of us who are thus compelled! :)
I believe it is as Clive said, to detect the presence of someone waiting to cross. When the request to cross button is pressed the crossing sequence starts, it takes a few seconds. If the detector sees that you have gone, it will cancel the cross sequence. If you stand still at a working one, you can see the red led come on/go off
@@pineappleroad when I did a year in traffic signals, they trialled a ‘video’ detector that detected change in colour rather than IR, I don’t know if they continued using them though. They were cool bits of kit, with onboard intelligent processing, although I did have one defeated because it was full of condensation.
Often have these cancel on me and constant repressing can mean they never trigger, find the lights themselves to take much longer to change compared to other non-sensor lights. Often I can cross safely before the "man" turns green, annoyingly being 200 yards away and finding the system didn't cancel/notice I'd moved away, don't f######g like them at all, grrrr.
@@dav1dbone i find them annoying as well, the trick I’ve come up with to get them to behave is to sway back and forth a bit underneath the sensor, which, at least for me, is enough to prevent it from cancelling (although there is one crossing which has a bent pole where i have to find the right spot)
@@pineappleroad I've been doing that since they first appeared around here, the trick for me is moving in a non suspicious manner, lol. I firmly believe that this particular set of lights is defective though.
I believe they work a little differently to what you describe based on my observations. The detector cannot see people 'waiting', instead the detectors are watching the crossing itself for it being occupied. You will often see on a light controlled pedestrian crossing that isnt part of a road junction that the light will turn green just after the last crossing pedestrian has reached the pavement. There isnt a fixed time setting for the red to be shown.
There are some pedestrian crossings near where i live which do have sensors to detect when people are waiting (but you still have to press the button to activate the sequence) And if the crossing detects that no one is waiting, then it will cancel the pedestrian request to cross I have had it cancel on me when i was stood waiting to cross, a little too still, i ended up having to press the button again and wobble underneath the sensor
@@philgifford8258 ... that was what was puzzling me. Perhaps the unit can filter out transient car noise and thereby pick out slower occupants of the crossing area.
I think it could be part of a puffin crossing. There is no fixed time and no flashing green man. Traffic remains stopped until the sensor detects no more pedestrians on the crossing.
Your correct there, however it will have a time out period, which also means when these fail the crossing is still operational if you get what I mean, but it will yes be used for a puffin or toucan
They are very clever devices Clive, they also detect weather the pedestrian is walking quick or slow telling the lights that somebody is still crossing.
@@johndododoe1411 Yeah we have an amber man for “clearance” with green man being an “invitation to cross”. The amber man is 1 second for every 1.2 meters
Worked at a place in the late 90's that was huge and had motion and/or infrared detection everywhere to cover the grounds, (football field size) It was all solid state from the late 1960's and was heavy duty military spec. They decided to add to indoor detection also, (cheaper late 90's technology) and the smallish orange sized emitter detectors, did it all. One day I was carrying a 1-yard x 1-yard sheet of plexiglass in the door and noticed, that the light didn't blink... it needed the movement and the heat signature from my body, so the games began, as we left the sheet by the door, to see who the best at sneaking past it was.
Most of the traffic signalling devices I have encountered in my little area of Ontario, Canada doesn't seem to utilise these sensors. It's usually just a momentary switch or a capacitive sensor mounted on one of the poles that pedestrians push or touch to cross the street. One of the fancier units for the blind also has haptic feed back when you touch or push the button. It pulses briefly when you push it and does a subsonic vibration when the signal is on in that direction. It also has two tone beep or a chirp playing followed by "Walk sign is on." If you hold the button down, the loud speaker tells you the direction and which street. The only thing is these are quite complex and tend to malfunction or brake like the crosswalk LED signs themselves.
I've been finding the audible signals more and more often here in the other CA. It's a good idea - when properly executed. I recall at least one intersection played the locating signal (for blind people to follow across) _from all four corners._ More dangerous than helpful, I'd think.
Interesting to see a "Schrack" branded relais in there... one of the few remaining tech companies in Austria; they were also very big in telecommunicatino in the 1990s and next door to where I worked then.
The pulsing removes the effect of slow changing ambient IR (DC filter, with a capacitor) and the net high/low differential pulse output is what is detected and passed on for control.
Very interesting discrete implementation of this idea, I love the non-square grid layout of the emitter and sensor diodes, presumably it's designed to be placed at a very specific height and angle to the crossing. Those remind me of all the railway crossing protection equipment in Japan, those sorts of IR units are common along with simple beam break, LiDAR, radar and visual cameras designed to detect pedestrians, cars and trucks in the crossing when the gates are down. If they detect something they trigger warning lights well before the crossing and approaching trains immediately stop or continue slowly. With how often I see news stories of trains hitting high centered trucks on US railway crossing I don't understand why this technology is never implemented there, although I have seen a couple of those on UK rail crossings but I don't believe there is any requirement.
Just some quick notes: 1. Obstacle detection is not widespread since many crossings are not busy enough to warrant it, and four-quadrant gates seem to work well enough. 2. Many of the UK's level crossings are CCTV-monitored. Open crossings have floodlights and trains usually go somewhat slow through them. Overall, a pretty good connection, though!
The reason in the US is size. This place is fucking huge, and plenty of places have their own small local governments and councils that want to be the ones who had the idea in the first place, or it's not happening. If you wanted to make it universal to have those crossing protections, you would need to go through many thousands of petty governor's and mayor's.
Japan is Japan. They almost always obey all the rules. In America trains would have to stop on every crossing, because of people activating these sensors. At some point, train drivers would just ignore the warnings not to stop 15th time in a row...
Boy does it drive me nuts when people compare something going on in some other tiny country to the US. Things like, "wHy iSn'T tHeRe mOrE mAsS tRanSiT?" Also, Japan has a 99% criminal conviction rate. So people are really kept in line. They have the highest number of innocent people in prison in the first world. That's how they keep their population like robots.
There should be a Part 2 to this video where the analog receiver section gets reverse engineered. That ancient PIC has no ADC and all I see is opamps and comparators, plus the obvious two analog muxes. So there's something potentially interesting going on there.
So, it sounds like this is basically a somewhat upgraded, but very large and clunky, PIR detector. I was thinking it might be able to detect pedestrians in the crosswalk, and extend the light until the crosswalk actually clears, but it doesn't seem this is anywhere near that sophisticated.
No I think you’re right. We have a type of crossing called the “puffin crossing” (google uk road crossings, they all have names and are wonderfully British) which has these to cancel the signal if the human goes away, but will also extend the signal so slow people can get across. I think they’re like temporary traffic lights that can sense cars waiting and switch the lights appropriately instead of using a timer.
We use 2 detectors, an on crossing detector which tells the controller that pedestrians are crossing and a kerbside detector which monitors the area around where the pedestrians wait to cross.
@@johndododoe1411 he means traffic lights installed temporarily to cover road works. Permanent traffic lights usually (but not always) use ground loops for traffic detection.
A tidy PCB and plenty of output and input sensors to detect any movement in front of the unit. It is interesting to see items like this that we intent to wait on to do their job, without us thinking of how it is doing it.
Clive, I was standing at a pelican crossing one day when it was raining very heavy and a puddle was forming on the road beside the crossing. I kept taking a step back to avoid inconsiderate drivers soaking me and the traffic lights remained on Green. I felt the infra red was at the wrong angle otherwise the traffic would have stopped.
The front will be Perspex 962 or an equivalent. It's a black acrylic using a dye that just happens to be transparent to near-IR. Many years ago when I first used it they had a disclaimer saying as the IR transmission was a side effect it was not guaranteed.
A person entering and leaving the detection field leaves a trail. The IR LED emitters are narrow-pulsed so the detectors can differentiate pulsed IR from ambient background IR. The detection is synchronised, to completely cancel ambient signal. A bit like a 'lock in amplifier'. The design is vaguely similar to motor vehicle windscreen rain drop detector.
The array of detectors combined with the lens form a very crude imaging device (camera) which could, by turning on the lights one at a time, scan the space in front of it and determine by parallax how far away the "seen" surface is. With only a 5x6 pixel array though it isn't clear whether you would have much depth resolution.
FWIW, traffic light systems have massive failsafe design go into them. So sensor units will not cancel any part of a sequence - they just affect the timing. Someone still crossing; hold the timing longer (but no longer than the maximum crossing time.) No-one there, get back to green for traffic subject to minimum crossing time. Also, if the unit fails or gets stuck it must allow a normal ped crossing sequence to run
They had something similar to these throughout my Michigan 1981-built high school. It was a technologically advanced building (it had coaxial ethernet and digital clocks in 1981), a sprawling state-of-the-art 350,000sf campus in the middle of a field. They were used both for intrusion/security and occupancy detection for HVAC dampers. It may have been acoustic but I'm quite confident it was light based. I've seen the acoustic ones in buildings from the late '60s so the idea is well seasoned.
We have something similar to this at some railroad crossings. It’s a two part system to detect any tractor trailers/lorries for you lol. So the crossing arms don’t go down trapping them.
I worked with traffic detectors for over a decade, the unit places a "call" to the traffic controller upon a detection event, and that call is overwhelmingly in the form of a contact closure to ground, grounding that input into the traffic controller letting it know that there are pedestrians detected.
At least on traffic controllers in the USA, they can be configured to handle detector inputs two ways. "Pulse" where the detector outputs a single pulse for each detection. In this case the call is stored until that phase is serviced. Pedestrian push buttons and upstream (from the stop line) loop detectors are configured this way. The second way is that the detector output is continuous as long as a vehicle is detected. In this case the call is ONLY serviced if the detector is still sending an output. No output, the phase is skipped. This is used for stop-line detection. I've never seen any use of pedestrian detection in the USA (pedestrian push buttons are all I ever see), but it would probably be configured to skip the pedestrian phase if no detection output.
I always thought those buttons and sensors at the lights don't actually work and they're all just set on timers and the button is just to give pedestrians a sense of control.
I'm thinking that it's more of a shape detector looking for human shaped objects, then possibly the delay/bypass people that don't move. It may need to block learn or be put in learn mode to learn it's environment profile. Hard to say any for sure without the software. Thanks for the look!
Given the severity of the keystone, I would think this is a crossing as opposed to kerbside detector. I like our quirkiness of naming types of crossing after birds, with the pelican, toucan and puffin varieties, although the latter two are a little contrived. A lot of people still refer to "pelican crossings" tho.
I suspect it can not only detect the presence of an object, but also it’s direction, distance and speed. This way it will ignore perpendicular vehicle movements, but detect objects moving towards and away.
Interesting approach to the keystone correction. That must have taken quite some testing to know exactly where to place the bottom LEDs and the corresponding receivers. It looks quite simple at first glance but then the intricate complexity stands out. I love circuits like this where the board layout changes very subtly to accommodate how the environmental factors affect its operation. Great video!
Would've been nice to see some analysis of the signals coming from the unit under different conditions. Whoever did the PCB layout and enclosure design deserve cookies.
These work as a smart replacement for pedestrian light call buttons. I remember seeing these when visiting a neighboring country. I didn't know they existed back then and wondered: why do the traffic lightd go green as soon as I approach the crossing ?
Since the light is pulsed and matched with the receiver, I think you could filter out ambient noise by measuring the receiver with the source light off, then by measuring with it on, then the differential measurement will exclude ambient light and interference
On the later models of the HB Modules traffic radars we used varistors instead of fuses to protect the power supply and relay contacts against human error wiring them up wrong in the field.
Huh I was just expecting it to strobe in random waves and "listen" to both the background and its own light flood, turns out that it's a bit more complex
Theres a few crossings near me that have sensors to detect if people are waiting to cross (you still have to press the button, the sensors are there to detect if there are still people waiting to cross) and how it is supposed to work, is if the people waiting walk away or cross when there is a gap, it will cancel the request to cross And i have had it cancel on me when i have been stood waiting, a little too still When it cancels i then have to press the button again, and “wobble” underneath the sensor so that it doesn’t cancel again
@@nickwallette6201 That would be interesting. Just before I retirerd Iwas working on an automatic number plate reading traffic sign. It read the plate and flashed ithat and the speed pl;us a sad-happy face depending on your speed. I havent seen any so it might have been binned.
@@dcallan812 Image recognition .. I still can't comprehend how that works. Really cool stuff. We've come a long way from RealVideo playing postage stamp-sized video files at 15fps.
@@nickwallette6201 When you think what the first web cams were like compaired to 4K videos on a divice that fits in your pocket and has a screen made up of millions of pixels. 👍
Regarding Fresnel lenses. Such a useful invention. Widely used in lighthouses. They are so effective that during the day the lenses have to keep turning when the light is off in order that the sun does not get focused onto the same spot on the bulb too long lest it be burnt.
The sensor helps the system track pedestrians crossing the road so traffic isn’t held longer than necessary. Ie it goes to green once it knows you’ve crossed or giving extra time to say and elderly person.
Now that you've done the light mechanism, you should do a follow up on the switch and speaker. As we push the crossing switch, here in Alameda County, CA, US, the speaker says "wait" very loudly ( usually in a recorded man's voice, sometimes bass sometimes more tenor). As the light turns you hear, "walk on the crosswalk Broadway Avenue". Have someone disassemble and send the entire street package! Also kudos to your donor of the lamp and electronics, it must have taken quite the postage to mail!
have you looked at the IR pulse? if the pulse is a multiple pulse or some form of modulation then that modulation can be picked up by the receivers and that helps distinguish a real target from a random IR% signal. Andy
Near my workplace the crossings will remain through to traffic until such time that they detect a gap in the traffic large enough for a vehicle to stop safely, then activate the traffic light sequence.
Somehow, I feel that you would probably assemble a pelican crossing in your house at somepoint, just because of the fun of having such a thing that most people just take for granted... :P
Great timing! I just discovered a small, handheld box amongst things I bought at auction. Turns out it is a homemade 9v battery powered IR detector. Now I can aim it at crosswalk sensors to see what's going on.
It would be interesting to look at, while working, with a video camera which does not have an IR filter. One might guess that the circuit uses a corrorlateor between the send and receive signals. If the mark space ratio of the transmit LEDs provide space for the receiver photodiodes to measure the ambient background light level, then it would not be difficult to compensate for it. I came across a person detector that.used two photo detectors to look at a common area, through two separate lens, it then amplified the differential of the two resulting signals. The effect was that short term changes in light level, produced by a shadow crossing the field of view were detected, while long term light level changes were ignored. A later version of a similar detector used the circuit from an optical mouse to detect changes in an image focused on the optical mouse light detector.
All the crossings where I live appear to have "press the button" to activate the crossing. There are some traffic lights that use a microwave head to detect traffic approaching the lights to activate the change of traffic lights. Somewhere, I have a microwave Doppler head that is used for the traffic lights. If I can find the microwave Doppler head, I'll see if it is worth sending for "investigation". K Watt.
Yes, we have press the button crossings as well. In some places, the button press doesn't actually do anything other than give people a button to press. In other places, it will start the crossing sequence, but if you cross the road before it shows the green man, if for example there is a natural gap in the traffic, then it will cancel the sequence.
Doppler is the standard vehicle detector mounted facing traffic flow. Used in conjunction with induction loops to set the extension timers for the flow of traffic.
Interesting. Here in the US, all we got are (sometimes placebo) buttons you push to trigger the crosswalk signal. We don't got anything fancy like automatic pedestrian detectors, at least not around my area.
Kinda wish they did have pedestrian detectors in the USA, there are quite a few people that will push the button and then won't wait for the crossing signal. So then a whole bunch of traffic gets to wait a minute for a pedestrian signal for a pedestrian that crossed long before the light turned red.
I'm sure these are used to detect if there's pedestrians still on the road so that the traffic signals remain at red if there's a particularly slow person crossing, at least the ones I've seen in the wild do that. My assumption is that the relay would switch a signal coming from the main controller for the junction/crossing.
I can't help thinking the 'Human Seagull' is going to have problems with this on his way home from the pub... Does it detect half man - half Seagull creatures that are staggering? 🤣 Thank god they're not fitted to Pelican crossings... Seagulls and Pelicans... They have some history!
If that circuit board was designed in 1997 and those chips made in 2001 (which matches the look of that overall design) then it definitely predates LED moonflowers as 5mm colored LEDs which project didn't exist back then - the best you had was some dim indicators. If that particular unit saw any real use then those LEDs are most likely shot - even pulsed, IR LEDs don't last that long, especially if you run high current through them.
High brightness LED's did exist in 2000, they were quite expensive for the green and blue, those were made by Nichia. Red, orange and Yellow high brightness LED's were available from Toshiba.
Microchip changed their PIC controllers from "C" to "F" in the late 1990's. Chips like the 16C84 became the 16F84, even though the chips themselves did not change internally. This was used in marketing so that Microchip Technology could emphasize their use of "Flash"memory.
I thought the C's were one time programmable as I remember before the days of Arduino using the EPROM versions of the PIC controllers to debug code before finally burning a C part. These days it's too easy - no need for a nasty, smelly UV lamp when you have a bug in your code!
@@jameswest848 I remember wasting quite a lot of the C parts due to stupid bugs. I seem to remember that the erasable windowed versions were horribly expensive
I think that out of all of the C parts, only the C84 was re-programmable. That's the reason it became the 'hobbiest chip of choice'. Otherwise you had to devolop with the MPLAB in circuit emulator (very expensive) or UV erasable ceramic versions of the chips.
That'd be red LEDs, since those Infrared VCSEL diodes already on this sensor have nearly the same forward voltage, only off by a few hundred millivolts so it would work since the current management is done by resistors in matrix switch circuits. That would be interesting to see it that way after some modifications.
Hi, Clive if you used your phone's camera it would allow you to see the infrared scanning of the emitting LEDs. I was told if you are ever found that the traffic lights have locked on a red signal get out of the car and press the crossing signal button and it will trigger an interrupt on the MCU. A traffic light tech told me this little gem of info and the system goes back to the normal working state. I don't if that is still the case but it is still worth trying if you are in that position in the future.
@@allancopland1768 Hi Allan I am aware the DSLR is not designed for IR but someone told me this trick. when people have complained that remote control is not working. I just grab my cellphone and check to see if it is working 9 times out of 10 they are working and the only problem is a dirty lens on the remote an easy fix for the customer.
Nice device, high quality and interesting pattern of LEDs and photodiodes. Seemingly irregular, but this irregularity almost certainly serves some purpose.
I've only seen it cancel the crossing once, many years ago when the system was first implemented. Lattery, either by bad configuration or deliberate design I have never seen it cancel from a 'walk away'.
They often get disabled, as it can be an annoyance and safety issue for pedestrians if it cancels at the wrong times. There's never any markings on the pavements as to where the detection zones are, so it's very easy for pedestrians to inadvertently walk out of the detection zone.
i would check the leds with a cro, normally these divices feed the ir led a tone then look for the same tone on the reciever, phase shift can be used to calculate distance and allows it to rule out all ambiant ir light because it will only see the tone its feeding the ir led and ignores everything else, in an array like that it would be fairly accurate
Now I want to see the unit firing the beams. I always love this kind of stuff. Would be interesting to see the pattern created by the specific arrangement and specifically the "wave" it creates. Maybe replace it with visible spectrum LEDs for reference? Keep this "exotic" and industrial stuff coming. Always great to see what everyday stuff can do that we don't even think about.
In a large room, or outside at night, position a small, bright , light source behind a fresnel, approximately the distance it would be from the IR led's. Then with a piece of paper, see where the light comes to a focus. Then try this same thing, but in the position of the other two groups of LED's (there appears to be three different groups of LED's, arranged in three different density groups. The IR sensors are also arranged in three similar patterns.
Hard to believe loosely aimed LEDs & photodiodes are creating much of an image. Would have been useful to see the LED projection pattern & the diode readout pattern in a drawing.
Fascinating bit of electronica. I have cheapo little solar-rechargeable outdoor lights that use the same tech; an IR transmitter/sensor activates the light at night. I think it also senses solar IR during the day to prevent it lighting when not needed because it does not kick on. Either that, or it uses voltage from the solar cell to tell the MC to not power on the light.
Many old burglar alarms used this system. I worked for Chubb Alarms division in the 80’s and sold many of these. Today we use passive infrared detectors which are tiny compared to this hulk.
I was going to ask if the led's fired separately. Because it would create perfect stereoscopic images. But it doesn't look like it actually has a camera, just uv sensors. But I am only halfway through the video. You might get into it more.
There's a safety requirement in IEC 60825-1 that under any foreseeable single fault, the light output can't exceed eye-safe intensity limits. It's an issue particularly with IR LEDs and lasers because you don't have the natural aversion to staring at a bright light. The microcontroller duty limiting circuit is very likely intended to meet this. You'll find similar design considerations and current limiting in LED and laser mice.
Not to mention doing it that way also have a nice side effect; derating any of the semiconductor light emitters also ensure that they last much longer too.
what about sharks with lazer beamz on their fricking heads?
@@Palmit_ 🤣
I doubt a pure software solution will meet that "functional safety" standard. Thus there should be a design aspect that stops runaway light output even if the software microcontroller runs amok. Maybe the switching transistors have a pulse width limiting driver circuit, or maybe they rely on frying the LEDs or resistors before they fry anyone's eyes.
The lazer mice will fix it no worries
Heehee! I used to work at Microsense! (I was a programmer working on their comms, monitoring systems, and control, rather than detection, which was quite specialised). We were always a little bit of a minnow around some much bigger sharks, but it meant we often got 'interesting' requests from customers (mostly local authorities) that the bigger companies didn't have the flexibility for. One of our higher ups on the team was heavily into optics, and did some great fun work on both detection and signals, improving over the older systems as LED technology started to take over.
There were several AIM detectors in the suite, with the LEDs configured for roadside, gantry, kerbside and stop line use.
Telent own Microsense now don't they? - very intresting insight BTW 👍
@@samverve Via Traffic Signals UK, and Traffic Signals EU (who sequentially took on the MSL brand first). I was there for the end of MSL, all of the above, and tiny a bit of telent, before moving away and up to Glasgow. Over the years we got up to some fascinating stuff in signal, detection and control.
And my enduring memory is the internal part number for a waterproof toggle switch cover for the controller manual panels, which was ME-NIPPLE. :)
@@moogthedog2816 haha, it's always great when someone gets away with sneaking a good name in somewhere.
Always liked the Microsense kit. Along with AGD and Peek.
Also worked with ferranti and gec marconi always impressed with how many edge connector repairs could be done on the old ferranti boards.
Fun trivia:
The term 'Pelican crossing' comes from PeLiCon - a contraction of 'Pedestrian Light Controlled'. This type of crossing has two fixed time periods: a steady green man / red light to vehicles (indicating pedestrians may cross / vehicles must wait) followed by a flashing green man / flashing amber light to vehicles (indicating pedestrians may continue to cross, but must not start to cross / vehicles may proceed if the crossing is clear).
The term 'Puffin crossing' comes from PUFIN - a contraction of 'Pedestrian User-Friendly Intelligent'. This type of crossing does away with the second fixed period (i.e. the flashing green man / amber) and instead uses the sensor, shown in the video, to detect if pedestrians are still on the crossing - if so, it extends the initial steady green man phase until the crossing is clear.
I've heard of a 'pelican crossing' and a 'zebra crossing' but never a 'puffin crossing'.
@@th3w1gg3r9 They are quite similar to British railroad crossing signals.
@@douro20 As in Jungle Book...I heard a diamond ring but I never seen an elephant fly?
Don’t forget the TOUCAN crossing which is a combined pedestrian/cycle crossing.
@@douro20 If you're in the UK, puffins are the ones where the signal that it's safe to cross is annoyingly placed in the same unit as the call button, instead of being on the pole on the opposite side of the street.
Clive, the inner 'infrastructure nerd' in me loves things like this - it's a long shot but would love to see what's inside a speed camera. Definitely love to see more street furniture and infrastructure!
Brings back memories. Used to work on traffic lights years ago. There's loads of cool tech in traffic. Doppler radar, video detection, crossing tech to help vision impaired people, emergency vehicle priority transponders.
Good old AIM-K detector. Used to work for Microsense and these were a really good kerbside detector for use at Puffin Crossings. These self learn on power up and you keep the area clear so they learn what a normal clear crossing looks like. This one looks like a later unit manufactured in Hungary by videoton the same people that make televisions and other electrical products
Hi Clive, when you said "People don't stay still at lights" is why I kinda think I know how these things work, usually you can see the red LED that shows detection as a pedestrian (in the ones I've spent time looking at!). If you're a wheelchair user and you don't move, it will cancel when you've pressed the call button, which is quite annoying. I'm pretty sure they use these to stop people running up, hitting the button and stopping traffic.... so annoying kids basically, but yeah, now I have to wiggle when sat waiting at lights or it sometimes places me in the 'background decoration' category :p
I guess the detection is done by lock-in amplification:
When the LED is on, the ADC value from the sensor is added to a variable. When it´s off, the value is subtracted. This is done for a couple of strobes.
This eliminates external light disturbances and indirectly amplifies the light of the LEDs that an object is reflecting.
Changes in the magnitude of the reflection can be easily registered.
Correct, there are 2 types of detectors, On-crossing and Kerbside. As you can probably guess, the Kerbside dets looks for peds at the kerbside that have pressed the wait button, if the det no longer detects peds at the crossing it can cancel the wait button demand (as its called) . The On-crossing detectors look at the crossing itself and will hold the lights on green (for the pedestrians) up to to the max green time if it detects pedestrians still on the crossing eg lots of people crossing, elderly crossig slowly etc.
This is genius, I don't think we have anything like this used in the U.S. It's all just pretimed from what I can gather. Hit the button, it eventually gives you a crossing signal, some have a countdown clock, others will begin to flash the pedestrian symbol about 10 seconds before it turns back. I like the idea that the system can adjust for slow or fast walkers.
Definitely could use one of those auto-cancel features in many ped/bike crossings here. People just push the button, ran through the red anyway, and let 15 cars sitting through a minute at red.
@@greentjmtl there are a number of crossings near me that have sensors to detect if there is still someone waiting, unfortunately, the sensors aren’t perfect, if you are stood too still, they don’t see you, quite a few times ive been stood waiting at one, and it has cancelled, so i end up having to wobble underneath the sensor
@@pineappleroad And then there's the impossible task of designing safe and fair control algorithms so people don't wait unnecessarily or get killed.
@@johndododoe1411 Its called MOVA by Siemens
That makes me want to replace the emitter LEDs with alternating red, yellow, green, blue, then set it up with its lens and film it at high speed.
👋🏻 Hi chris
Just use an IR camera that can see the existing colors and pattern.
I just love these careful, clever analyses by BigClive. The hardware design is brilliant - especially with the special dispersion pattern of the LEDs and IR receivers - but Clive hit the nail on the head several times, especially with the "magic" software, which IMHO is reminiscent of advanced digital video synthesis on earlier radar systems, which wrestle with the same detection and measurement problems in terms of determining what is ground clutter and filtering it out in a smart way without losing the ability of detecting and tracking objects "hiding" in the clutter.
When it comes to hiding the clutter, it somewhat reminds me of those detectors for automatic door controlls, those usually learn as well to workl somehwat predicatably. otherwise even having for example a large dark floor mat on a lighter tile or floor (which doesn´t seem to be a rare instance) could set them of. If you are patient enough and able to stand quite still you can test it, if you remain very still in front of them, at some point they will stop onening and closing and remain closed.
I remember that we had a unit just like this (or highly similar at least) in the late 70s as part of a very high-end home alarm system.
The chips might not have been a part of it, might all have been done with discrete parts though...
The ir detectors used to be added to junctions that had issues with cyclists and motorbikes not being able to trigger the induction loops due to the alloys used to make them. By pointing them just in front of the stop line they allowed them to trigger the junction and get a green light.
Doppler radar works ok for cars due to a flat reflective surface but causes issues with the reflected wave of a bike
They've been using video detection for this purpose for about last 25 years here (Virginia, USA). The first system they used was the Econolite Autoscope system. It had the video processor in the same enclosure as the camera. It used an X86 CPU, think it was a 386. There was a broken one at a job I worked at, but I never got a chance to tear it down to see what was inside.
@@brianleeper5737 The video detectors came into use around the time I left the industry. It was a while ago.
The Infrared emitters in this pedestrian sensor may actually be a VCSEL diode rather than a LED, as I took a look at the emitters where you got the camera close enough to the board, I could deduct that it is a laser diode due to the fact that the VCSEL diode die looks fundamentally different than the LEDs when you look into the LED package, that is with the electrically connected laser optical cavity in the center of the die (and the wire bonds tend to be welded at the corner of VCSEL die). It could be a 940nm Infrared VCSEL, who knows?
they do look odd and being labeled as "LD#" suggests to me they're laser also.
also, it seems the camera does not pick them up at all. even when he mentioned a glimmer, it looked like nothing but reflected ambient lights.
@@throttlebottle5906 Human eye can detect pulsed IR light up into the 1300nm (iirc), while the hot mirrors on normal cameras will reject almost all 850nm and almost almost all 950nm already. If these are 1060nm or 1100nm VCSELs, then it'd be probably entirely invisible to most cameras but the eye can still sense it.
@@-szega Yes. That would be the correct assumption, unless you have a camera that lets 940nm Infrared light through for purpose of night vision.
Another fun fact about crossings is that they are all named after birds. Pelican, puffin and toucan meaning 2 can cross pedestrians and cycles. Also Pegasus which is for horses but also has wings to fit with the bird theme. A cuckoo was a term used to describe a control system from another manufacturer fitted in the wrong case.
Never heard of a crossing having a name..... I must live in the wrong country.
Wow thanks never heard of the others..most just say zebra crossing occasionally pelican..Wats the puffin then
It's Two Can crossing for ¶issed paedestrians😵
@@foxxy46213 bit of a stretched acronym. pedestrian user friendly intelligent. A puffin puts the red/green indicators above the button and makes you look in the direction of traffic to see them.
I recently learned about Hawk signals/crossings as well, contributing to the bird theme. There were a bunch in Tucson, AZ
The marking around the infrared LEDs uses the abbreviation “LD“ instead of LED. Usually when I see that… It stands for “laser diode“. Which would make a lot more sense because they would be much more focused and are dirt cheap these days.
I would think maybe it was originally designed for low power infrared laser diodes but they found they could get away with LEDs. Just a thought.
Also the photo diodes are properly labeled and abbreviated “PD“.
I’ve seen similar set up an alignment used in those laser tire alignment machines used in automotive shops. I’ve had to repair a few of those for random clients and they use almost the exact same setup just a bit more advanced.
Yep, I agree, it's a laser diode, since I already can tell, having handled both regular LEDs and VCSELs in the same 5mm LED epoxy packaging (they look the same except when you take a look at the die in it through the lens, it's going to be obvious here, they look different).
It is looking for the change in the pulse signals from the emitter. By looking for the pulsed signal it is not sensitive to slowly changing IR from sunlight.
It is now relatively easy to spot Big Clive viewers. They will be the ones staying visible at Pelican crossings by dancing.
A pedestrian crossing here in UK has a motor underneath with a plastic knob on.
It’s for blind people to hold once it starts to turn the light is red
And that is not as simple as you think as it has to withstand being stalled.
I never conceived that they might be "smart" in that sense. Thank you Clive.
I remember back in the 90s... one particular crossing would give false negatives all the time... I took to jiggling around at that crossing until they got the calibration sussed out.
I think that rather than cancelling the call, they actually come into play once the crossing has changed in favour of the pedestrian. It holds off the change back to traffic mode until all the pedestrians are clear in order to aid less able users. They also allow quicker return to traffic mode if an athlete scurries across. In general, once a pedestrian has pushed the button, the lights will cycle regardless of whether the user waits or not. The scallywags round here used to stick chewing gum in the button and leave the lights to keep changing, showing that there is no call cancelling.
The LEDs may be modulated to discriminate it's own signal from steady state IR sources (like sunlight). This is a very common technique, even consumer IR remote controls do this.
I remember at a long-past former employer we had a person tracker device to allow a pan/tilt camera to keep the person centered. The person being monitored (in this case a patient who was also connected to our EEG equipment) would wear IR emitters. The emitters were modulated and this mostly worked well to keep the camera from locking onto the window when the patient moved past it. Until some of the hospitals that our equipment was used in changed their fluorescent lights to new ones that coincidentally had the same (or similar enough) frequency as the modulated IR that we used for patient tracking. Had to do a quick design change to change the modulation frequency that we used in every unit that was affected.
I still remember when I was a kid,standing outside the Royal Albion Hotel in Brighton,in the middle of the beam controlling the traffic lights. I still remember my Dad coming around the corner,too!
Hello from North Carolina in the US, Clive. I'm Eric another Friend of Dorothy, and fan of your interesting explorative and expansive videos -- and now your PINK CALCULATOR also!
I will try not to be jealous, kind sir, just as you should avoid falling under the spell of the green-eyed monster when I tell you about my Pink, Crystal Studded HDMI cable, quite possibly the gayest HDMI cable found on any of the inner planets.
My background is in software (more specifically, software PROCESS and Design Engineering) so I lack a lot of basic electronics understanding and eduction but I still find your videos very interesting and informative. Please do keep on doing what you do. Your videos are compelling for those of us who are thus compelled! :)
I believe it is as Clive said, to detect the presence of someone waiting to cross. When the request to cross button is pressed the crossing sequence starts, it takes a few seconds. If the detector sees that you have gone, it will cancel the cross sequence. If you stand still at a working one, you can see the red led come on/go off
Quite a few times ive had one cancel on me because i was stood too still (i end up having to wobble to keep it from cancelling)
@@pineappleroad when I did a year in traffic signals, they trialled a ‘video’ detector that detected change in colour rather than IR, I don’t know if they continued using them though. They were cool bits of kit, with onboard intelligent processing, although I did have one defeated because it was full of condensation.
Often have these cancel on me and constant repressing can mean they never trigger, find the lights themselves to take much longer to change compared to other non-sensor lights. Often I can cross safely before the "man" turns green, annoyingly being 200 yards away and finding the system didn't cancel/notice I'd moved away, don't f######g like them at all, grrrr.
@@dav1dbone i find them annoying as well, the trick I’ve come up with to get them to behave is to sway back and forth a bit underneath the sensor, which, at least for me, is enough to prevent it from cancelling (although there is one crossing which has a bent pole where i have to find the right spot)
@@pineappleroad I've been doing that since they first appeared around here, the trick for me is moving in a non suspicious manner, lol. I firmly believe that this particular set of lights is defective though.
I believe they work a little differently to what you describe based on my observations. The detector cannot see people 'waiting', instead the detectors are watching the crossing itself for it being occupied. You will often see on a light controlled pedestrian crossing that isnt part of a road junction that the light will turn green just after the last crossing pedestrian has reached the pavement. There isnt a fixed time setting for the red to be shown.
Exactly what I was thinking. I've seen them on crossings situated at the top of the pole and pointing across the crossing, one on either side.
There are some pedestrian crossings near where i live which do have sensors to detect when people are waiting (but you still have to press the button to activate the sequence)
And if the crossing detects that no one is waiting, then it will cancel the pedestrian request to cross
I have had it cancel on me when i was stood waiting to cross, a little too still, i ended up having to press the button again and wobble underneath the sensor
@@pineappleroad Pedestrians must wobble to be noticed, nice!
Can the sensor differentiate between pedestrians moving & traffic crossing at different angle?
@@philgifford8258 ... that was what was puzzling me. Perhaps the unit can filter out transient car noise and thereby pick out slower occupants of the crossing area.
I think it could be part of a puffin crossing. There is no fixed time and no flashing green man. Traffic remains stopped until the sensor detects no more pedestrians on the crossing.
Your correct there, however it will have a time out period, which also means when these fail the crossing is still operational if you get what I mean, but it will yes be used for a puffin or toucan
They are very clever devices Clive, they also detect weather the pedestrian is walking quick or slow telling the lights that somebody is still crossing.
so, basically it's a very low-res IR camera with some image recognition. Cool!
They’re used for call cancel and in Ireland specifically for extending the amber man time as people cross
You have yellow man lights? We only have red and green men.
@@johndododoe1411 Yeah we have an amber man for “clearance” with green man being an “invitation to cross”. The amber man is 1 second for every 1.2 meters
Worked at a place in the late 90's that was huge and had motion and/or infrared detection everywhere to cover the grounds, (football field size) It was all solid state from the late 1960's and was heavy duty military spec. They decided to add to indoor detection also, (cheaper late 90's technology) and the smallish orange sized emitter detectors, did it all. One day I was carrying a 1-yard x 1-yard sheet of plexiglass in the door and noticed, that the light didn't blink... it needed the movement and the heat signature from my body, so the games began, as we left the sheet by the door, to see who the best at sneaking past it was.
Most of the traffic signalling devices I have encountered in my little area of Ontario, Canada doesn't seem to utilise these sensors. It's usually just a momentary switch or a capacitive sensor mounted on one of the poles that pedestrians push or touch to cross the street. One of the fancier units for the blind also has haptic feed back when you touch or push the button. It pulses briefly when you push it and does a subsonic vibration when the signal is on in that direction. It also has two tone beep or a chirp playing followed by "Walk sign is on." If you hold the button down, the loud speaker tells you the direction and which street. The only thing is these are quite complex and tend to malfunction or brake like the crosswalk LED signs themselves.
I've been finding the audible signals more and more often here in the other CA. It's a good idea - when properly executed.
I recall at least one intersection played the locating signal (for blind people to follow across) _from all four corners._ More dangerous than helpful, I'd think.
Interesting to see a "Schrack" branded relais in there... one of the few remaining tech companies in Austria; they were also very big in telecommunicatino in the 1990s and next door to where I worked then.
The name Schrack is still around, nowadays used by TE (Tyco Electronics) Connectivity.
That sensor would be going off crazy like around me. As you can tell, even my sense of humour is pedestrian
The pulsing removes the effect of slow changing ambient IR (DC filter, with a capacitor) and the net high/low differential pulse output is what is detected and passed on for control.
Very interesting discrete implementation of this idea, I love the non-square grid layout of the emitter and sensor diodes, presumably it's designed to be placed at a very specific height and angle to the crossing.
Those remind me of all the railway crossing protection equipment in Japan, those sorts of IR units are common along with simple beam break, LiDAR, radar and visual cameras designed to detect pedestrians, cars and trucks in the crossing when the gates are down. If they detect something they trigger warning lights well before the crossing and approaching trains immediately stop or continue slowly. With how often I see news stories of trains hitting high centered trucks on US railway crossing I don't understand why this technology is never implemented there, although I have seen a couple of those on UK rail crossings but I don't believe there is any requirement.
Just some quick notes:
1. Obstacle detection is not widespread since many crossings are not busy enough to warrant it, and four-quadrant gates seem to work well enough.
2. Many of the UK's level crossings are CCTV-monitored. Open crossings have floodlights and trains usually go somewhat slow through them.
Overall, a pretty good connection, though!
The reason in the US is size. This place is fucking huge, and plenty of places have their own small local governments and councils that want to be the ones who had the idea in the first place, or it's not happening.
If you wanted to make it universal to have those crossing protections, you would need to go through many thousands of petty governor's and mayor's.
Japan is Japan. They almost always obey all the rules. In America trains would have to stop on every crossing, because of people activating these sensors. At some point, train drivers would just ignore the warnings not to stop 15th time in a row...
Boy does it drive me nuts when people compare something going on in some other tiny country to the US. Things like, "wHy iSn'T tHeRe mOrE mAsS tRanSiT?" Also, Japan has a 99% criminal conviction rate. So people are really kept in line. They have the highest number of innocent people in prison in the first world. That's how they keep their population like robots.
It's not implemented here because our culture values the US Dollar above the human life.
There should be a Part 2 to this video where the analog receiver section gets reverse engineered. That ancient PIC has no ADC and all I see is opamps and comparators, plus the obvious two analog muxes. So there's something potentially interesting going on there.
So, it sounds like this is basically a somewhat upgraded, but very large and clunky, PIR detector. I was thinking it might be able to detect pedestrians in the crosswalk, and extend the light until the crosswalk actually clears, but it doesn't seem this is anywhere near that sophisticated.
No I think you’re right. We have a type of crossing called the “puffin crossing” (google uk road crossings, they all have names and are wonderfully British) which has these to cancel the signal if the human goes away, but will also extend the signal so slow people can get across.
I think they’re like temporary traffic lights that can sense cars waiting and switch the lights appropriately instead of using a timer.
@@ncot_tech Why temporary?
We use 2 detectors, an on crossing detector which tells the controller that pedestrians are crossing and a kerbside detector which monitors the area around where the pedestrians wait to cross.
Wouldn’t be passive IR with the emitters though, would it?
@@johndododoe1411 he means traffic lights installed temporarily to cover road works. Permanent traffic lights usually (but not always) use ground loops for traffic detection.
A tidy PCB and plenty of output and input sensors to detect any movement in front of the unit. It is interesting to see items like this that we intent to wait on to do their job, without us thinking of how it is doing it.
Clive, I was standing at a pelican crossing one day when it was raining very heavy and a puddle was forming on the road beside the crossing. I kept taking a step back to avoid inconsiderate drivers soaking me and the traffic lights remained on Green. I felt the infra red was at the wrong angle otherwise the traffic would have stopped.
The front will be Perspex 962 or an equivalent. It's a black acrylic using a dye that just happens to be transparent to near-IR. Many years ago when I first used it they had a disclaimer saying as the IR transmission was a side effect it was not guaranteed.
A person entering and leaving the detection field leaves a trail. The IR LED emitters are narrow-pulsed so the detectors can differentiate pulsed IR from ambient background IR. The detection is synchronised, to completely cancel ambient signal. A bit like a 'lock in amplifier'.
The design is vaguely similar to motor vehicle windscreen rain drop detector.
The array of detectors combined with the lens form a very crude imaging device (camera) which could, by turning on the lights one at a time, scan the space in front of it and determine by parallax how far away the "seen" surface is. With only a 5x6 pixel array though it isn't clear whether you would have much depth resolution.
Basically my thoughts. It's a really low resolution camera.
FWIW, traffic light systems have massive failsafe design go into them. So sensor units will not cancel any part of a sequence - they just affect the timing. Someone still crossing; hold the timing longer (but no longer than the maximum crossing time.) No-one there, get back to green for traffic subject to minimum crossing time.
Also, if the unit fails or gets stuck it must allow a normal ped crossing sequence to run
They had something similar to these throughout my Michigan 1981-built high school. It was a technologically advanced building (it had coaxial ethernet and digital clocks in 1981), a sprawling state-of-the-art 350,000sf campus in the middle of a field. They were used both for intrusion/security and occupancy detection for HVAC dampers. It may have been acoustic but I'm quite confident it was light based. I've seen the acoustic ones in buildings from the late '60s so the idea is well seasoned.
Clive, the two 1 ohm resisters @ 2:20 are in parallel not series, if in series they would be 2 ohms.
We have something similar to this at some railroad crossings. It’s a two part system to detect any tractor trailers/lorries for you lol. So the crossing arms don’t go down trapping them.
I worked with traffic detectors for over a decade, the unit places a "call" to the traffic controller upon a detection event, and that call is overwhelmingly in the form of a contact closure to ground, grounding that input into the traffic controller letting it know that there are pedestrians detected.
At least on traffic controllers in the USA, they can be configured to handle detector inputs two ways. "Pulse" where the detector outputs a single pulse for each detection. In this case the call is stored until that phase is serviced. Pedestrian push buttons and upstream (from the stop line) loop detectors are configured this way. The second way is that the detector output is continuous as long as a vehicle is detected. In this case the call is ONLY serviced if the detector is still sending an output. No output, the phase is skipped. This is used for stop-line detection. I've never seen any use of pedestrian detection in the USA (pedestrian push buttons are all I ever see), but it would probably be configured to skip the pedestrian phase if no detection output.
I always thought those buttons and sensors at the lights don't actually work and they're all just set on timers and the button is just to give pedestrians a sense of control.
It's called a "beg button".
I'm thinking that it's more of a shape detector looking for human shaped objects, then possibly the delay/bypass people that don't move. It may need to block learn or be put in learn mode to learn it's environment profile. Hard to say any for sure without the software.
Thanks for the look!
Given the severity of the keystone, I would think this is a crossing as opposed to kerbside detector.
I like our quirkiness of naming types of crossing after birds, with the pelican, toucan and puffin varieties, although the latter two are a little contrived. A lot of people still refer to "pelican crossings" tho.
I suspect it can not only detect the presence of an object, but also it’s direction, distance and speed. This way it will ignore perpendicular vehicle movements, but detect objects moving towards and away.
Interesting approach to the keystone correction. That must have taken quite some testing to know exactly where to place the bottom LEDs and the corresponding receivers. It looks quite simple at first glance but then the intricate complexity stands out. I love circuits like this where the board layout changes very subtly to accommodate how the environmental factors affect its operation. Great video!
Would've been nice to see some analysis of the signals coming from the unit under different conditions.
Whoever did the PCB layout and enclosure design deserve cookies.
These work as a smart replacement for pedestrian light call buttons. I remember seeing these when visiting a neighboring country. I didn't know they existed back then and wondered: why do the traffic lightd go green as soon as I approach the crossing ?
Since the light is pulsed and matched with the receiver, I think you could filter out ambient noise by measuring the receiver with the source light off, then by measuring with it on, then the differential measurement will exclude ambient light and interference
On the later models of the HB Modules traffic radars we used varistors instead of fuses to protect the power supply and relay contacts against human error wiring them up wrong in the field.
PTC thermistors? They can protect against intermittent faults and people "upgrading" fuses.
@@bigclivedotcom Yes, that's what I meant. Brain cells misfiring after large brandy
@@Jon6429 Good choice of tipple.
Huh
I was just expecting it to strobe in random waves and "listen" to both the background and its own light flood, turns out that it's a bit more complex
Theres a few crossings near me that have sensors to detect if people are waiting to cross (you still have to press the button, the sensors are there to detect if there are still people waiting to cross) and how it is supposed to work, is if the people waiting walk away or cross when there is a gap, it will cancel the request to cross
And i have had it cancel on me when i have been stood waiting, a little too still
When it cancels i then have to press the button again, and “wobble” underneath the sensor so that it doesn’t cancel again
The shape of the LEDs and reciver hardware is an unusual pattern. I wonder what the beam looks like.Great bit of kit 2x👍
It would be really neat to get a serial feed of the sensor intensity and plot it as blocks of pixels in a grid. :-)
@@nickwallette6201 That would be interesting. Just before I retirerd Iwas working on an automatic number plate reading traffic sign. It read the plate and flashed ithat and the speed pl;us a sad-happy face depending on your speed. I havent seen any so it might have been binned.
@@dcallan812 Image recognition .. I still can't comprehend how that works. Really cool stuff. We've come a long way from RealVideo playing postage stamp-sized video files at 15fps.
@@nickwallette6201 When you think what the first web cams were like compaired to 4K videos on a divice that fits in your pocket and has a screen made up of millions of pixels. 👍
Regarding Fresnel lenses. Such a useful invention. Widely used in lighthouses. They are so effective that during the day the lenses have to keep turning when the light is off in order that the sun does not get focused onto the same spot on the bulb too long lest it be burnt.
The sensor helps the system track pedestrians crossing the road so traffic isn’t held longer than necessary. Ie it goes to green once it knows you’ve crossed or giving extra time to say and elderly person.
Very cool to see the matching arrays of LEDs and sensors. Would like to see scope traces of output and input.
Now that you've done the light mechanism, you should do a follow up on the switch and speaker. As we push the crossing switch, here in Alameda County, CA, US, the speaker says "wait" very loudly ( usually in a recorded man's voice, sometimes bass sometimes more tenor). As the light turns you hear, "walk on the crosswalk Broadway Avenue". Have someone disassemble and send the entire street package!
Also kudos to your donor of the lamp and electronics, it must have taken quite the postage to mail!
have you looked at the IR pulse? if the pulse is a multiple pulse or some form of modulation then that modulation can be picked up by the receivers and that helps distinguish a real target from a random IR% signal.
Andy
You should look at it through an IR camera to look at the beam shape and distance ?
We're using lidar now in Austin to track pedestrian crossings and to lengthen the interval 👏
Near my workplace the crossings will remain through to traffic until such time that they detect a gap in the traffic large enough for a vehicle to stop safely, then activate the traffic light sequence.
I wonder what happened to that pedestrian crossing..
Somehow, I feel that you would probably assemble a pelican crossing in your house at somepoint, just because of the fun of having such a thing that most people just take for granted... :P
Great timing! I just discovered a small, handheld box amongst things I bought at auction. Turns out it is a homemade 9v battery powered IR detector. Now I can aim it at crosswalk sensors to see what's going on.
It would be interesting to look at, while working, with a video camera which does not have an IR filter. One might guess that the circuit uses a corrorlateor between the send and receive signals. If the mark space ratio of the transmit LEDs provide space for the receiver photodiodes to measure the ambient background light level, then it would not be difficult to compensate for it.
I came across a person detector that.used two photo detectors to look at a common area, through two separate lens, it then amplified the differential of the two resulting signals. The effect was that short term changes in light level, produced by a shadow crossing the field of view were detected, while long term light level changes were ignored. A later version of a similar detector used the circuit from an optical mouse to detect changes in an image focused on the optical mouse light detector.
All the crossings where I live appear to have "press the button" to activate the crossing.
There are some traffic lights that use a microwave head to detect traffic approaching the lights to activate the change of traffic lights.
Somewhere, I have a microwave Doppler head that is used for the traffic lights.
If I can find the microwave Doppler head, I'll see if it is worth sending for "investigation".
K Watt.
Yes, we have press the button crossings as well. In some places, the button press doesn't actually do anything other than give people a button to press. In other places, it will start the crossing sequence, but if you cross the road before it shows the green man, if for example there is a natural gap in the traffic, then it will cancel the sequence.
Doppler is the standard vehicle detector mounted facing traffic flow. Used in conjunction with induction loops to set the extension timers for the flow of traffic.
Interesting. Here in the US, all we got are (sometimes placebo) buttons you push to trigger the crosswalk signal. We don't got anything fancy like automatic pedestrian detectors, at least not around my area.
Kinda wish they did have pedestrian detectors in the USA, there are quite a few people that will push the button and then won't wait for the crossing signal. So then a whole bunch of traffic gets to wait a minute for a pedestrian signal for a pedestrian that crossed long before the light turned red.
I'm sure these are used to detect if there's pedestrians still on the road so that the traffic signals remain at red if there's a particularly slow person crossing, at least the ones I've seen in the wild do that. My assumption is that the relay would switch a signal coming from the main controller for the junction/crossing.
Looks like the most uncomfortable VR headset yet
I can't help thinking the 'Human Seagull' is going to have problems with this on his way home from the pub... Does it detect half man - half Seagull creatures that are staggering? 🤣
Thank god they're not fitted to Pelican crossings... Seagulls and Pelicans... They have some history!
Sweet tear down. I didn't even know they had detectors on cross walks.
Some of them have a spinney thing under the button you press for blind or partially sighted people.
If that circuit board was designed in 1997 and those chips made in 2001 (which matches the look of that overall design) then it definitely predates LED moonflowers as 5mm colored LEDs which project didn't exist back then - the best you had was some dim indicators. If that particular unit saw any real use then those LEDs are most likely shot - even pulsed, IR LEDs don't last that long, especially if you run high current through them.
IrDA emitters were pulsed at 0.5A - but only for 1.63us. However they are (were?) for intermittent use.
High brightness LED's did exist in 2000, they were quite expensive for the green and blue, those were made by Nichia. Red, orange and Yellow high brightness LED's were available from Toshiba.
Microchip changed their PIC controllers from "C" to "F" in the late 1990's.
Chips like the 16C84 became the 16F84, even though the chips themselves did not change internally.
This was used in marketing so that Microchip Technology could emphasize their use of "Flash"memory.
I thought the C's were one time programmable as I remember before the days of Arduino using the EPROM versions of the PIC controllers to debug code before finally burning a C part. These days it's too easy - no need for a nasty, smelly UV lamp when you have a bug in your code!
Actually they did change as the C version is one time programmable and F version is reprogrammable.
I thought the C chips were write once for final production?
@@jameswest848 I remember wasting quite a lot of the C parts due to stupid bugs. I seem to remember that the erasable windowed versions were horribly expensive
I think that out of all of the C parts, only the C84 was re-programmable. That's the reason it became the 'hobbiest chip of choice'. Otherwise you had to devolop with the MPLAB in circuit emulator (very expensive) or UV erasable ceramic versions of the chips.
Interesting. Would be nice to replace the LEDs with 'normal' LEDs to see what beam patern is outputed..
That'd be red LEDs, since those Infrared VCSEL diodes already on this sensor have nearly the same forward voltage, only off by a few hundred millivolts so it would work since the current management is done by resistors in matrix switch circuits. That would be interesting to see it that way after some modifications.
Hi, Clive if you used your phone's camera it would allow you to see the infrared scanning of the emitting LEDs.
I was told if you are ever found that the traffic lights have locked on a red signal get out of the car and press the crossing signal button and it will trigger an interrupt on the MCU. A traffic light tech told me this little gem of info and the system goes back to the normal working state. I don't if that is still the case but it is still worth trying if you are in that position in the future.
Most DSLRs see pretty well in the IR although they are really not intended for that use.
@@allancopland1768 Hi Allan I am aware the DSLR is not designed for IR but someone told me this trick. when people have complained that remote control is not working. I just grab my cellphone and check to see if it is working 9 times out of 10 they are working and the only problem is a dirty lens on the remote an easy fix for the customer.
@@davidgrey943 yes, I checked my cellphone camera and it can see the IR from my TV remote.
Nice device, high quality and interesting pattern of LEDs and photodiodes. Seemingly irregular, but this irregularity almost certainly serves some purpose.
The cink palculator catches me off guard every time
I've only seen it cancel the crossing once, many years ago when the system was first implemented. Lattery, either by bad configuration or deliberate design I have never seen it cancel from a 'walk away'.
They often get disabled, as it can be an annoyance and safety issue for pedestrians if it cancels at the wrong times. There's never any markings on the pavements as to where the detection zones are, so it's very easy for pedestrians to inadvertently walk out of the detection zone.
i would check the leds with a cro, normally these divices feed the ir led a tone then look for the same tone on the reciever, phase shift can be used to calculate distance and allows it to rule out all ambiant ir light because it will only see the tone its feeding the ir led and ignores everything else, in an array like that it would be fairly accurate
That was very interesting! I've seen these things next to pedestrian lights in NL and always wondered what it was. I think there's also sonar ones!!
Much more complex than expected.
Now I want to see the unit firing the beams. I always love this kind of stuff. Would be interesting to see the pattern created by the specific arrangement and specifically the "wave" it creates.
Maybe replace it with visible spectrum LEDs for reference?
Keep this "exotic" and industrial stuff coming. Always great to see what everyday stuff can do that we don't even think about.
I read _pedestrian_ in the title as the adjective rather than the noun, and was briefly puzzled.
In a large room, or outside at night, position a small, bright , light source behind a fresnel, approximately the distance it would be from the IR led's. Then with a piece of paper, see where the light comes to a focus. Then try this same thing, but in the position of the other two groups of LED's (there appears to be three different groups of LED's, arranged in three different density groups. The IR sensors are also arranged in three similar patterns.
Hard to believe loosely aimed LEDs & photodiodes are creating much of an image. Would have been useful to see the LED projection pattern & the diode readout pattern in a drawing.
Fascinating bit of electronica. I have cheapo little solar-rechargeable outdoor lights that use the same tech; an IR transmitter/sensor activates the light at night. I think it also senses solar IR during the day to prevent it lighting when not needed because it does not kick on. Either that, or it uses voltage from the solar cell to tell the MC to not power on the light.
Many old burglar alarms used this system. I worked for Chubb Alarms division in the 80’s and sold many of these. Today we use passive infrared detectors which are tiny compared to this hulk.
Thanks Clive looks fab a lot of work goes into them by the looks of things
This human-detector looks perfect for installation on a Skynet Hunter-Killer.
its essentially a massive version of what is in a phone.
I was going to ask if the led's fired separately. Because it would create perfect stereoscopic images. But it doesn't look like it actually has a camera, just uv sensors. But I am only halfway through the video. You might get into it more.