[Sticky] Introduce Yourself
i am arlette from switzerland and i am not a programmer, nor a technical nerd but an artist interested in arduino and sensors. specificly i am trying to find cheap sensors similay to laser because lasers, which are not fix installed in box are forbidden in switzerland since summer 2019. now i just watched the super tutorial about lidar and was trying it out with the tfmini. the first try did not work. then i found out, that the wires on my tfmini are different. the white one is the rx and the green one on the right side is the tx. now i switched them and bingo - it works - perfect. but it is not really the thing i try to find. as i mentioned above i am searching a sensor similar to a laster pointer because i would like to detect the place of a person in a room which triggers a specific function. the whole thing should also work when several people are in the room. this means, the focus should be on one person but it should be possible to change the focus with a movement. it is a bit complicated i know. the idea with the lasers was, that there would be a row of red dots on the wall, sendet from an object in the middle of the room. if a person moves the connection would be interrupted an something is going to change at the object in the middle of the room. but unfortunately this idea does not work anymore because the open sending lasers are forbidden.
another idea was to use a kinect which i bought a few years ago. unfortunately i don't understand this thing and i don't know how to get it work. the problem seems to be that it need something like a driver for my mac. i have no idea how to find out, what i have to do. difficult without having any knowlede...
maybe someone has a good idea, how i could get it done? unfortunately i will not be the good member which can give some good advices because i am just working with tutorials and try and error and i am free of mathematical talent ;o)
i hope you welcome my anyway
1) Rotate Lidar sensor. If not possible due to laws on lasers consider other possibilities.
2) Rotate an ultrasonic sensor and measure the distance to the reflection every so many degrees. There are some with horns and housings which can operate up to about 7 meters. See Adafruit Large Ultrasonic (Sonar) Sensor with Horn and UART Output PRODUCT ID: 4664 Arduino has one on their catalog that says 10 meters.
3) Rotate thermal image sensor. This is nice because the temperature will tell you whether you have a living person unless you have a very warm room. Adafruit AMG8833 is one possibility.
4) Rotate a tv camera and use AI to decide if it is people. This can be very inexpensive but is probably most software complex. I put vision on a Raspberry PI with a servo for under $20 US. I did not attempt any AI software with it. Arduino offers a Portenta product with an available vision shield that is targeting AI applications. Not so inexpensive. But I wonder if there might be sketches developed sometime in the future. Just a guess.
If you don't make a full rotation you can get by with flexible wiring, but if full rotation is required you would need slip rings.
I have not really any experience with any of the above but lots of old experience in using sensors for various purposes.
I have a very shallow understanding of circuits/electricity, but am trying to learn. I am comfortable with computers and technology in general. I run Linux on my laptop, have a Pi-hole running on my network etc. I am currently trying to learn more about networking, servers, and Arduino/Maker. I have a SparkFun Arduino Uno clone, a bunch of sensors, a Raspberry Pi Pico, an Arduino Pro Mini (I think that is what it is called...5V version) and a smattering of other bits and bobbles. I still need a multimeter, but despite much reading and watching on The Interwebs choosing a meter that will not end in buyers remorse is proving quite difficult. A Fluke is out of the question, but the $20 Amazon/eBay jobs always seems to be lacking something another article or video says I will need. Even the $60 meters are confusing. Hoping to start in a few weeks.
For work I practically have to use my Fluke but at home as a hobbyist, I use the EEVBlog branded BM235(It is just a rebranded Brymen but they are a fairly good brand). I love it and I had no buyers remorse, and that is unusual for me. There are more options out now but it really does what I need and it was not that expensive. I also preferred to rely on the knowledge of someone I trust in electronics.
Here is a spreadsheet of multimeters. It can be daunting but there is all the facts in one place.
But more importantly, even with the knowledge that a $10 meter will do everything I need, I don't regret it. It is easy to use and see.
I am not completely clear as to what you are trying to do, but I am visualising it is somewhat similar to the 'laser' beams sometimes shown in the movies to detect intruders that the 'hero' must avoid breaking. Except that in your case, you intend the people to occasionally break a beam to initiate a specific function.
And you also say "the focus should be on one person but it should be possible to change the focus with a movement", which I am wondering if that means that the system should only respond if a particular person breaks the beam... or maybe I am misunderstanding your intentions. How you would implement this I have no idea. ... presumably a very smart image recognition system?
So with apologies for any misconceptions on my part, I think it would be helpful if you clarified what you are expecting the detection system to achieve.
I am not familiar with Swiss laws but a quick Google suggested that the ban was specific to laser pointers of the type sometimes used in lecture rooms, etc., unless they were labelled and classified as Class 1 (the lowest power category).
Laser light sources above Class 1 are by definition at least hazardous, and looking directly into even a Class 1 laser can be very unpleasant. Hence, they are not suitable for situations in which someone might look directly into the beam (or a reflection of the beam), so even without such a law, I think your idea needs modification.
If you are looking to detect breaking a beam based on simple light source and detector, then maybe consider the infra red LED and photodiode pairs used for TV remote controls. These work by pulsing the beam and then detecting the pulses, thereby cancelling out most of the ambient light sources. The light levels are much lower and naturally 'eye safe'. Depending upon the distance (which will be more limited than a laser could achieve) and how 'narrow' you need the beam to be, you will need to read the specs of the LEDs and photodiodes, and consider adding lenses and/or tubes at both ends. Of course a simple system like this cannot differentiate what or who breaks the beam. Furthermore, the beam width will be much wider than the 1mm or so often achieved by lasers, which may or may not make it useless for your intentions.
Hi all. I'm Dave from the UK. I have virtually no experience with electronics and only get a microcontroller working by copying exactly someone else's work, so I'm looking for a better understanding so that I can be a little more creative.
I'm a network engineer working on WANs and more recently software-defined WANs, so if I can help anyone out in that respect just holler.
I'd like to thank Bill for keeping me sane during the last couple of years. When it seemed that the world was falling apart, there was always a Dronebot Workshop video to restore some serenity!
Lastly, bragging rights to anyone who knows the source of my username without the use of Google!
hi larry - thank you so much for your inputs.
my tfmini seams to be one, that only «scans» one point. the one you suggest would be a 360 degree one? the one that is presentet in the workshop «Getting Started with LIDAR» ? - since my know-how is very low i am not sure if i should buy one of those things - just to fail...
i made some tests with the «normal» distance sensors. i hope you can give me some advise to my problems with the ultrasonic-sensors an ir-sensors.
i tried a hc-sr04. as long as i test it with something with a hard cover everything is works good. but it doesn't really function well with dark clothes. if i check it on serial monitor on the arduino it produces lots of mistakes when there is a fluffy material to be detected. that means, that the big one you suggest probabely won't work better.
the same or similar problem do i have with the ir-sensors.
the idea with thermal image sensor sounds good to me. i will try it out.
and the idea with the portenta sounds interesting to... thanks for this - i will try it out to.
but this will take its time - and a bit of money...;o)
in the meantime i was making some tests with an old kinect. but i have some problems because it is an old kinect 1414 version 1. i could get it run with processing but now i do not know how i can get it together with arduino...
and what happens, if there are more then one person in the room...?
thanks for your help
hi dave - thank you for your answer. my former idea was, that there is an object in the middle of a room throwing laserpoints to the walls around it. if a person interrupts the laserlight - something happens with the object in the middle of the room. therefore i wanted to by about 100 simlpe laser modules for very less money. the idea existed before switzerland did forbid laser. unfortunately i could not find any class 1 laser modules to use for my idea. and also i wasn't aware that to detect a possible interruption of the laser i would need a detector on the other side. what i am trying to find is something that can throw a light to the wall an at the same time can detect where someone is standing. the throwing light could be faked but the detection of a person in a room is quite complicated without specific knowledge. i was trying a tfmini and anold kinect - both are not the solution yet... maybe you have a good idea?
Wow what a spreadsheet! Thanks!
In the US I have been using EXTECH Typically in the $30 to $60 range for about 6 years. They aren't the Flukes I used to use when working for big corporations, but they serve me well. I usually like to have two meters so if something seems weird I can get a different check on it.
Especially true on those extremely rare occasions where I care about True RMS. Then I would like at least 2 different models of meters and preferably from different manufacturers. Though today who knows how many meters come from the same company in China.
My next multimeter may likely be an Amazon Commercial. No idea who actually makes them and whether Amazon participated in the design. My guess is no, but anymore who knows.
Hi arlette .. @goldfisch,
I fear you are setting yourself a difficult target to meet, but maybe others can come up with some suggestions.
As you have realised, with light (and to some extent ultrasound), there are two obvious 'simple' detection schemes.
The first being when the source and detector are essentially co-linear facing each other, so the 'beam' is incident upon the detector until something (including someone) interrupts it. This method usually provides a longer range capability and tends to be more specific in the range of positions that it will detect an object.
The second being when the source and detector both point into the detection area, with the principle that an object entering the area will reflect some of the 'beam' back onto the detector. In this case it is common, for convenience but not mandatory, for the source and detector to be very close to each other. When the source and detector are co-located, then only one location needs to be wired to, etc. However, many surfaces, particularly common fabrics used for clothing, tend to absorb both light and sound, so reflections are commonly weak. Perhaps worse, their reflectivity with light especially, may vary by orders of magnitude, between say a dark coloured coat and a white shirt, so the strength of received signal will present more difficulties. I am not personally familiar with TFmini, which you mention, but a quick Google suggests it is in this category, using infra red light source and detector mounted in a small package.
In addition to light and sound, radar is starting to become available as it is being developed for applications such as self driving cars and robots, but I don't know if you will find any 'off-the-shelf' solutions.
Of course, with both of the above approaches, any 'sizeable object' that can move into the detection area will trigger a response. By careful selection of detection area and sensitivity, it may be possible to limit it to a 'whole' person detection, as against a small object or just a hand wave, but tracking a specific person when there are several persons likely to enter and leave the detection area would not normally be feasible.
A different, but probably impractical, approach is to provide each person with a 'token' that can be detected. RFID tokens being an obvious starting point for consideration.
You also mention Microsoft Kinect. Again, I haven't any personal experience but my understanding is that MS have developed algorithms for recognising the 'mechanical' aspects of a person (arms, legs, etc.) with regard to interpreting what the person is doing. Again, a quick web query confirmed that Microsoft Kinect itself has effectively been 'retired' for sometime, although it has been transformed into 'Azure Kinect', which includes a 'developer kit' for US $399.
Personally, I would be very wary of spending too much effort on an MS 'retired' product, unless you can find an independent 'support' community with the capability and enthusiasm of growing and maintaining it. I don't know if one exists but I haven't noticed one yet.
The MS Asure Kinect website says Azure Kinect can be used with any or 'no' cloud, as well as Azure, so a 'standalone' approach (probably needing a 'high end PC'?) should be possible,. However, I also noted that an FAQ said "The Azure Kinect DK is available for purchase in the United States, China, Germany, Japan and the United Kingdom.", so I don't know if that would be a barrier to you. I haven't looked closely, but my impression is that it is not aimed at recognising 'individuals' (in the 'face recognition' sense, as against the '2-legged person recognition' sense), so whether it will meet your other requirements is unclear. It should be good at recognising 'gestures' (possibly with limbs, not just facial expressions, etc.) that people make. Would that be of any interest?
Clearly, there are many other groups developing face and related image recognition systems .... Xilinx systems based on FPGA developments being one that comes to mind. However, these are often expensive and technically complex at present, although no doubt that will change as AI/ML methods move into the 'mainstream'.
You also mention Arduino. Whilst this is technically the name of an innovative company, not a product, it usually implies relatively low performance microcontrollers that are great for many control applications, but lack the number crunching ability implied for complex image processing. Obviously, these are 'soft' distinctions, but the implication is you are likely to need some relatively powerful processing capability to analyse the scene, (especially if it involves following more than one person at a time,) and then maybe send commands to an Arduino or similar to effect the response by enabling lights, motors, etc. as appropriate. Interfacing the image processing system to an Arduino system should generally be feasible using available network connections (e.g. Ethernet). The bigger challenge is likely to be the image analysis task to make the appropriate decisions.
So sorry if I don't have any good or original ideas ... I suggest you continue to think and discuss what you are trying to do ... maybe there is a 'smart' idea that isn't quite such a difficult technical (and maybe cost) challenge.
Good luck .... keep looking. Dave
Not exactly sure what you are trying to do but if it is to locate people then a LIDAR should work fine?
Unfortunately I don't have one to experiment with or have the software to use it as input to my own code tailored for particular applications.
You can see Bill moving about in the LIDAR example below.
In reference to using LIDAR, I did have a play way back when Bills video came out. In my limited play, because the LIDAR works on a very thin beam of light, unless a mechanism to 'wobble' the LIDAR is used it wont pick up much to recognise the persons outline to distinguish them from the background. If you are like me maybe a well rounded shape in a background of square walls may give a clue. 😎
I think a better bet would be a camera, and something like openCV to better pick out the presence of a person, and maybe even recognise their fizzog.
The problem with any solution is: Who writes the software for the specific purpose of detecting people and their positions as goldfisch indicated he was not a programmer.
With regards to LIDAR people move and thus would reveal themselves as people as opposed to walls and furniture. Software would have to be written to make use of that fact. Also maybe data from two or more LIDAR devices could be used for better resolution.
The problem with vision is that it doesn't really return the positional precision of LIDAR.
Maybe if there are high ceilings then downward looking cameras would work.
This weekend I'm on holiday (sounds weird to be "on holiday", as i am retired) at my sister's place in Errington, B.C., we grew up in Qualicum Beach. I'm excited because i found a banjo to fix up, ar a pawn shop in Parksville.
In my computer room / electronics lab i have been updating my Arduino based PCB for controlling a reflow oven, the newest version using primarily SMT devices on a Chinese manufactured board. The original control board, which i made 2011 or so, was etched "in house", as it were. I am also playing about with I2C implementation of displays, and with infrared remote control of NeoPixels ( not entirely successfully, i suck at the Arduino IDE)
I have temporarily set aside work on my CNC, and other guitar related shop activities
Enjoy your mini-vacation. You'll soon be stringing your family along with the banjo 🙂
Sion, the Unexpected Maker also made a reflow controller for a toaster oven and made several improvements to the first version. He's since graduated to a professional PnP and oven though.
I've been making plotters because I thought they'd be a challenge and they are 🙂 I've also made a bunch of useful things (row counters for knitters and other hobbyists, A4988 tuners, PCB boards and so on). I can't stop now as I have too many parts left in my inventory.
I also started a few projects of things like several versions of harmonographs, but instead programmed them on the Mac so I could use larger dimension arms and pendula.