Mini quad dectector gate

[Robert Longbottom]

Hi all,

After the mini-quad flying I was thinking of how to make a detector gate that would pickup a mini quad.  I don't think this is quite as easy as the usual "timing gate" problem where you can just have a single IR LED and detector because you could easily be just too high or low flying through to block the signal.  So I was thinking maybe a laser pointer and a light dependent resistor setup so that the laser beam bounces round inside a 3d printed hoop, possible helped by some tinfoil "mirrors", eventually coming to rest on the LDR.  Something like below.  I'm not sure if the LDR would work well or if there is something better that could be used - can you get LDRs that only detect laser light??  or some kind of filter?

We could probably do a number of different things with them

 - use two as some timing gates

 - connect up some LEDs and the LED changes colour each time you fly through, the aim being the person to get it onto green wins. (say)

 - have a counter with a display, 13th person through wins - think tactics...  Hang back and wait until its on 11 or 12 before going in or get in early and try to just go backwards / forwards to count up and block everyone else :-)

Thoughts?

Rob.

[Damian Axford]

Nice :)

thoughts:

• Use cheap mirror tiles on printed brackets attached to a squarish wood/aluminium frame - fast and easy to build, plus can quickly go larger than a typical build area
• Laser dioes are super cheap, can run off shared 5V supply (or battery) with an Arduino and work well with LDRs, however, they are visible light only so no good for most filters...  probably best bet is to add a snoot (tube) to the receiver to block out most of the ambient light - that plus a bit of tuning of the detector sensitivity should do.  You could use a simple red gel filter to cut out more ambient light (cut up old 3D glasses, etc).
• How about a neopixel indicator across the bottom that supports the various modes you suggested.... for counting games, it could work like a bar graph :)

[Malcolm Pett]

it woul be interesting to try a few different ideas.

[jmeosbn]

This looks like it should work, though the (albeit, more fiddly) mirror approach has better chance of success.  That said, you could always just try various emitter/detector combinations with one return signal from tin foil etc., then expand to 2,3..

[Robert Longbottom]

Cheap mirror tiles could work very well, if you can get tiny ones ( like 1cm square ), I was thinking tinfoil for ultra cheap / hacky.

I wasn't aware of laser diodes, will have to have a look into those. Do you know if you need some kind of "driver" board (I see you can buy such things) or can you just stick 5V across them?

I have some red transparency stuff courtesy of astronomy kit and also some baader (sp) film (I think) which is used in making solar filters and cuts out 99% of light, so I did wonder if that would cut visible light, but might let laser frequencies through.

The other idea I've had is if you had a number of detectors in a row you'd have to fly through them all and "light up" each one.

Some experimentation is required...

[Damian Axford]

laser diodes are what's inside every laser pointer :)  the low power ones are generally DC, 3-6V ish and don't need a driver board - go for a couple of quid on ebay, or hack apart a laser pointer.

lasers, by their nature, emit a very pure/tight frequency, so if the laser is visible, then there isn't any invisible component.  As a result, you need a suitably tuned visible light filter to work properly - or a crappy badly tuned thing that's vaguely in the right colour band.   gut feel is the red transparency stuff will probably work fine, the baader film (because it's an ND filter, not spectra specific) probably won't be much help as it'll just attenuate all light, rather than help isolate the laser light.  Mind you - filters are very quick to test, so no harm trying.

detectors in a row = cool :)

[jmeosbn .]

I'm sure I've seen smaller mirrors tiles - or something close - but probably sourced from a craft store.

I was thinking that one flying challenge could be along the floor under a certain height (12 inches or so) - or within a certain zone higher up. The multiple detector idea could work well for this..

Jamie Osborne

--
You received this message because you are subscribed to the Google Groups "swindon-hackspace" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swindon-hacksp...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Robert Longbottom]

Yeah see if you can get through without tripping any detectors.

[Robert Longbottom]

Do we have any LDR's at the hackspace?  I dont seem to have any at home :-(

If so I might bring in my laser colimater tomorrow to see what I can get to work.

[Jamie Osborne]

I don't recall seeing any specifically, but you could see if any of those robot kit parts use one?

Jamie

[Damian Axford]

I also have some IR reflectance sensors - might be possible to arrange them in a grid/circle and detect a copter going past....   would be very quick to prototype, so will try to remember them for wednesday.  Do we have a hot glue gun at the hackspace? 

[Robert Longbottom]

Cool.  Yes, I think we do have a glue gun.

I've ordered some laser diodes and some LDRs which will hopefully be here by Wednesday.  Do you know if you need to use a resistor with a laser diode (lime with and led)?

I've not really done much googling on them, but I keep seeing people referring to laser diode drivers?

On 19 January 2015 10:04:56 Damian Axford <dam...@axford.me.uk> wrote:

I also have some IR reflectance sensors - might be possible to arrange them in a grid/circle and detect a copter going past....   would be very quick to prototype, so will try to remember them for wednesday.  Do we have a hot glue gun at the hackspace? 

--

[Damian Axford]

Excellent... And generally it's just a DC voltage, drivers are only normally required for high power laser diodes.

[Robert Longbottom]

Yeah, I kind of get that impression. There are many links about using an
LM317 voltage regulator. Some of which refer to voltage and current
limiting. I can see how the LM317 regulates voltage, but not current (it's
just a variable voltage regulator as far as I can see). I'm thinking that
most of there links are for people trying to power them off batteries where
the voltage could be >5V.

It sounds like a resistor might be a good idea, but maybe I'll just try it
on the bench power supply with the current limiter on low to start with.
Just don't want to burn it out or burn a hole in the wall :-)

On 19 January 2015 11:50:12 Damian Axford <dam...@axford.me.uk> wrote:

> Excellent... And generally it's just a DC voltage, drivers are only
> normally required for high power laser diodes.
>

[Robert Longbottom]

The "laser"s (*) have arrived and they appear to already have a built in
resistor :-). Which makes me think I can just apply 5v.

Also I seem to have received one more than I ordered :-)

(*) in a Dr Evil (not Mr!) voice

[Damian Axford]

"excellent" (Burns style) - this may also be of use:

http://hackaday.com/2015/01/19/diy-hot-wheels-drag-race-timer/

Includes a laser diode timing gate!

[Robert Longbottom]

--

[Robert Longbottom]

Yeah, thats quite cool.  He does have a distinct advantage that his cars only really move in one dimension though :-)

I was wondering if it would be practical to stick an RFID tag to a quadcopter (light, uses no power) and so long as you pass "close enough" to the detector have it pick you up - you could use the "laser" to detect presence and the read the RFID tag to work out who it was!

[jmeosbn .]

Like the RFID idea! Things to check: how quickly can it be read; what tolerance for the angle presented.

I'd think the latter wouldn't be too much of an issue for RFID, but the former may be an issue when mounted to quickly moving quadcopters!

Jamie Osborne

--

[Robert Longbottom]

Based on the door cards at work angle doesn't matter.  But yeah read speed and distance from sensor could be an issue. 

Though maybe that's a different challenge, fly up to the sensor and hover close enough for long enough to activate it and light up your led.  Have a few of them spread around the room.  First person to light up all their LEDs wins :-)

[Damian Axford]

loving the: "fly up to the sensor and hover close enough for long enough to activate it and light up your led.  Have a few of them spread around the room.  First person to light up all their LEDs wins :-)" idea...  although RFID systems aren't the cheapest thing to play with - the readers especially are quite expensive.

[Robert Longbottom]

The build begineth :-)

[Jamie Osborne]

Looks like a good start (a little further and you would have been through!).

Jamie

On 24 Jan 2015, at 19:28, Robert Longbottom <Robe...@iname.com> wrote:

The build begineth :-)

<IMG_20150124_191626.jpg>

[Robert Longbottom]

Looks like it should work, though I suspect positioning and getting the mirrors to stay in place and resist knocks might need to involve some got glue! (The wood is already a bit chewed up where I've been flying through :-) )

Laser is top left, ldr is top right.  Pla glows in a really weird sparkly way under laser light.

On 24 January 2015 19:28:33 Robert Longbottom <Robe...@iname.com> wrote:

The build begineth :-)

--

[Jamie Osborne]

Looks awesome. Could you recess them; ideally from the outside? (Dunno if there are circular mirror tiles, but that would make things easier.)

Jamie

On 25 Jan 2015, at 11:47, Robert Longbottom <Robe...@iname.com> wrote:

Looks like it should work, though I suspect positioning and getting the mirrors to stay in place and resist knocks might need to involve some got glue! (The wood is already a bit chewed up where I've been flying through :-) )

Laser is top left, ldr is top right.  Pla glows in a really weird sparkly way under laser light.

<IMG_20150125_114029.jpg>
<IMG_20150125_114135.jpg>

On 24 January 2015 19:28:33 Robert Longbottom <Robe...@iname.com> wrote:

The build begineth :-)

<IMG_20150124_191626.jpg>

--
You received this message because you are subscribed to the Google Groups "swindon-hackspace" group.
To unsubscribe from this group and stop receiving emails from it, send an email to swindon-hacksp...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Robert Longbottom]

Yeah, recessing them could work, though you still need a way to fasten them in place and the right place so it can resist knocks and the beam follows the right path.  I'm thinking people going barrelling into the frame and knocking the whole thing over as well!  :-)

They had circular mirrors in hobbycraft which is where I got the square ones from, but I didn't look at the price compared to square.

[Jamie Osborne]

If the circular ones aren't much smaller (in area) then a counterbore from the outside followed by hot glue should work nicely both for position and (some) protection. :)

Jamie

[Robert Longbottom]

Made a bit more progress on this last night.  The mirrors, "laser" and detector are now hot-glued in place :-). And I extended the cabling so it can connect to a nearby arduino.  A short sketch of about 10lines of code (if that) shows that it works nicely and detects things passing through, reporting on the serial console for now.

Next steps are to add some LEDs, a buzzer, a counter, some other indicator, not sure what yet!

[Damian Axford]

awesome :)

[jmeosbn .]

Idea for this or any other quad obstacle: add a space heater to create tricky air currents for extra difficulty. These could be timed to turn on and off, or maybe triggered by another gate?

Jamie Osborne

On Thu, Jan 29, 2015 at 11:45 AM, Damian Axford <dam...@axford.me.uk> wrote:

awesome :)

[Robert Longbottom]

Yeah, good idea, I like the idea that you could fly somewhere else to activate it and blow someone else off course.  Or if you wanted a non-evil version, it could be on most of the time, protecting a goal and you have to go elsewhere to turn it off for (say) 10 seconds, to give you change to go and achieve the goal.  Very Crystal Maze!

I think I'd just use some old PC fans rather than a heater though.

[jmeosbn .]

I like both those ideas - PC fans sounds good, we'll have to wire some >= 80mm fans up and try!

Jamie Osborne

[Robert Longbottom]

So last nights efforts in this involved trying out a piezo  and another random buzzer I have to make some noise when someone goes through the gate.  Turns out the piezo was waaay to quiet, but the other buzzer was pretty good.  Most people seemed to be doing okay getting through the gate with their Hubsans and it took a fair old bashing with no real damage.  Some laser realignment was required a few times after it was completely knocked over, and there are quite a few battle scars on the wooden parts from propellers.  But all in all it survived okay.

Short video here: https://www.youtube.com/watch?v=T8enpxFL2ro 

[Jamie Osborne]

Also, perhaps a sensor activated by copter downdraft ?

Jamie

On 5 Feb 2015, at 13:37, Robert Longbottom <robe...@iname.com> wrote:

So last nights efforts in this involved trying out a piezo  and another random buzzer I have to make some noise when someone goes through the gate.  Turns out the piezo was waaay to quiet, but the other buzzer was pretty good.  Most people seemed to be doing okay getting through the gate with their Hubsans and it took a fair old bashing with no real damage.  Some laser realignment was required a few times after it was completely knocked over, and there are quite a few battle scars on the wooden parts from propellers.  But all in all it survived okay.

Short video here: https://www.youtube.com/watch?v=T8enpxFL2ro 

--

[Damian Axford]

a barometric pressure sensor could work for down-draft...  they have a fast sample rate and can detect fairly small pressure changes.  I'll try to rem to bring one Wed to test.

I also fancy trying an ultrasonic ping sensor, arranged something like this:

I suspect the two main issues will be:

• Getting the sample rate high enough to detect fast moving drones - http://playground.arduino.cc/Code/NewPing suggests a max rate of 30 per second, which after some fag packet maths, suggests a maximum speed of 6 miles per hour for micro drones, and probably closer to 15 miles per hour for Hubsans
• Somehow masking the sensor so that the detection zone isn't so conical... key being to get it as shallow as possible.  Of course, the more conical it is, the larger the detection area and the faster the drones can go :)

will also try to rem an ultrasonic sensor on Wed

[Robert Longbottom]

The barometric pressure sensor would work for down draft if its similar to the ones used on the quadcopter flight controller boards.  They are super sensitive.  You might need some kind of air flow funnel, or I'd imagine just flying somewhere vaguely near would be enough to set the thing off.  Also, advantage to people with more bigger / more powerful miniquads that create more down-draft :-)

I'd be interested to see how you get on with the ultra-sonic sensor.  I'm not sure how good they are at detecting small objects.  I've only ever really seen them used for detecting things like walls where there is a big "reflective" surface.

[Gyrobot]

Or if out doors, the wind might be a nuisance in more ways than one, playing havoc with both the sensor and the quads.

But I like the idea of enclosing the sensor inside a baffle box to reduce erroneous signals, and the gate becomes less physical and more "virtual" in a 3D printed physical sense, am I making sense? - NO!!

[Robert Longbottom]

erm...  If a quadcopter passes over the sensor in the woods, but the sensor doesn't detect it and noone sees it did the quadcopter actually cause a downdraft or was it using the force to levitate?  Thats waaay too philosophical for me :-)

On 25/02/2015 10:31, Gyrobot wrote:

Or if out doors, the wind might be a nuisance in more ways than one, playing havoc with both the sensor and the quads.

But I like the idea of enclosing the sensor inside a baffle box to reduce erroneous signals, and the gate becomes less physical and more "virtual" in a 3D printed physical sense, am I making sense? - NO!!

[Robert Longbottom]

Now available on a thingiverse near you :-)

http://www.thingiverse.com/thing:698512

[Jamie Osborne]

Ah! Good timing; we named a repo tonight to collect together each type of quad .. obstacle into!

https://github.com/snhack/quadstacle

Jamie

On 25 Feb 2015, at 20:41, Robert Longbottom <robe...@iname.com> wrote:

Now available on a thingiverse near you :-)

http://www.thingiverse.com/thing:698512

--

[Damian Axford]

excellent :)

prototyped last night:

1) Ultrasonic (HC-SR04) sensor 

This turned out to be utter crap...   

• detection was unreliable within 30cm
• at any range, required a large (at least A6) flat surface that was fairly perpendicular to the sensor
  
• was slow - maximal update rate was limited to 5-10 samples per second

Having configured it pointing vertically, it barely detected one flyby/hover out of many attempts.  

In summary - no good for a hover gate, and don't bother using HC-SR04 modules for much!

2) BMP085 pressure sensor

Turned out pretty well :)

• Sample rate set at 20 per second, could easily go faster
• Good detection range (with the addition of a paper funnel above the sensor) - from 0 to about 45cm worked
• Tracking changes in ambient pressure is necessary, as well as tuning a suitable detection range to avoid false positives

I made a breadboard test unit, that included a serial colour LED and a piezo buzzer.  The s/w (code on github) implements a basic hover game:

• When you hover over the sensor, the LED gets steadily brighter until it reaches a maximum value - at which point it flashes red and the buzzer beeps.  That equates to a couple of seconds hover time.
• If you drift off the sensor, the LED slowly fades - which gives you a chance to correct your hover and keep going

It's tricky, but doable - at least on a hubsan...  I totally failed to get there using the nano quad.

Next steps would be to turn it into an ATTiny-based stripboard circuit, add a battery pack and finish with a 3D printed enclosure (inc funnel)

[Robert Longbottom]

A good nights work by the sound of it.  I can't say I'm surprised about the sonar sensor being a bit rubbish.  No you say about the 30cm thing, I'm sure I remember reading they don't work so well for close distances.  Though 30cm isn't exactly close!

Pressure sensor pad sounds good though.  Hovering is very tricky with those little quads :-). I tried hovering in my gate to go back and forth and trigger multiple detections, no chance!  The pressure sensor approach could work quite well for bigger quads outdoors as well, so long as you can counteract the general windyness of being outside.

[Gyrobot]

 

Donald also brought in his RFID sensor and chips.

 

This would be a great way to go for coptor recognition and individual laptimes, stats etc. However with this Arduino version, the pickup antenna just wasn't sensitive enough for a flyby scenario, you would virtually/actually have to land on the thing and then take off. The antenna would have to be in a protective box which wouldn't be an issue. I think the kit was about £40 (?) a pop so it is not really a cheap alternative, it would be uber cool though maybe just for the start finish/line?

 

[jmeosbn .]

I've been wondering what to do with the camera the Hubsan h107C/D comes with, and also, the carrier "bay" and 8mm round aperture that the module usually sits in..

So, to throw an idea into the mix, how about one or more quadcopters replace their camera with a low voltage laser, while another (or more) detects* the laser and - optionally - cuts power* when "hit". So, basically, "Quadcopter LASER TAG!!!"

The only thing I'm thinking is that, with two moving points in 3D space, it may be near impossible to score a hit!

* Using the original camera would leave no space free (inside) for the beefy components needed to interpret the image; Rob's LDR based gate method is a lot less demanding on battery, price, and real estate/weight.

** This could be done gradually by faking out the voltage detection of the quad, though this might require extra components just to shunt the power through.

Jamie Osborne

On Fri, Feb 27, 2015 at 7:48 AM, Gyrobot <sp...@gyrobot.co.uk> wrote:

 

Donald also brought in his RFID sensor and chips.

 

This would be a great way to go for coptor recognition and individual laptimes, stats etc. However with this Arduino version, the pickup antenna just wasn't sensitive enough for a flyby scenario, you would virtually/actually have to land on the thing and then take off. The antenna would have to be in a protective box which wouldn't be an issue. I think the kit was about £40 (?) a pop so it is not really a cheap alternative, it would be uber cool though maybe just for the start finish/line?

 

--

[Gyrobot]

My eyes, my eyes!!!

But it sounds feasible. Another laser theme could be a fixed bullseye target with a detector in the middle that you have to "hit" with the onboard laser.

[Robert Longbottom]

:-). Yes nice idea, but I did think that might be a bit dodgy having lasers randomly shining around.  You'd have to be a pretty good pilot to get on target as well, unless you can find some way to spread out the laser beam or have a reasonably large diameter detector.  Maybe a printed "clear" cone would direct the light down to the ldr in the weird sparkly way.

On a similar vein, I wonder if an IR led pulsed at a set frequency could be used to detect particular copters (different people use different frequencies).  You'd have to overcome the problem of it reflecting off everything like ir remotes do, but maybe if you shielded the transmitter and detector in a similar way to either my gate or Damian's pressure sensor it could work.  We could also try running at reduced voltage to shorten the range.  Maybe more useful for a static target to fly up to that for shooting another copter out of the sky.

[Jamie Osborne]

We could all wear shades? ;)

Yes, I guess the accuracy of a laser is a disadvantage here - IR would be a good replacement if it can be focused enough (IR LEDs do have a specified angle of transmission/reception).

Identification can definitely be done. IR LED receivers already filter for a single carrier frequency - however, that frequency is then pulsed to overlay a data stream; or, in our case, continually pulsed (or perhaps repeating a pattern).

Jamie

[Damian Axford]

Beefy Quadstacle gate - mk1

Notes:

• Folds
• Supports both laser gate and hover detector
• Re-uses laser/LDR holders from Rob's design
• Hover detector cone connected to pressure sensor with silicone tube - keeps circuity at the bottom of the leg and away from props
• Ref below is based on 400mm lengths of 20x20 profile - probably too big for Hubsan size quads - it's parameterised, so easy to adjust to diff leg lengths
• Hubsan can just fit between the beams - but you'd have to be perfectly aligned and level as you went through the gate - almost impossible, so don't think it's an issue
• Legs (the vertical bars) held in position by bolted axles (bolt + wingnut)
• Toes held at max extension with string/fishing line - pinned to base of legs for stability
• Toes can be made longer if necessary - and/or the long printed section could be replace with a bit of dowel, alu tube, etc
• Blade protection to be added by wrapping legs/gantry in one of:
• Pipe insulation
• Foamcore
• Craft foam
• Colour scheme - always orange :)   with grey/black accents

[Damian Axford]

Added to git:

https://github.com/snhack/quadstacle/tree/master/BeefyGate/hardware

... assembly guide to follow, not that's really needed:

https://github.com/snhack/quadstacle/blob/master/BeefyGate/hardware/docs/QuadstacleGateAssemblyGuide.md

[Gyrobot]

Excellent, your OpenSCAD skills are a joy, just not your choice of colours, Gyrobot green with black wins all the time, just like Mint Aero beats Terry's Chocolate Orange.

[Damian Axford]

:)   and for the joy of parameterisation, I present the Gyrobot edition... of course, you'd also want a green laser

[Gyrobot]

Like +1

[Robert Longbottom]

This isn't Facebook!!!!  :-)

I'll be interested to see how well the folding works while keeping the laser aligned.

[Damian Axford]

Ditto... Esp. With that many bounces/mirrors :)

Can always make the folding brackets meatier...

[Damian Axford]

have printed a first set of parts, will test this evening, but for today's lunchtime randomness - I thought I'd have a crack at getting OpenCV + Python working on my Mac.  This is with a view to:

a) Having a script run after print jobs complete that emails me a webcam capture (using the @execute command in Repetier host) 

b) Detect QR codes (or similar) via webcam to support a super-bling Quad gate...     if this actually works, we can play with it on Wed

So.... 30mins to get OpenCV doing something, it all begins here:

http://www.jeffreythompson.org/blog/2013/08/22/update-installing-opencv-on-mac-mountain-lion/

[Robert Longbottom]

OpenCV + raspberry pi + webcam was pretty easy when I had a go with that.  Only a few lines of code to get a window onscreen with a live image.

Ill be interested to see if you can hover still enough for long enough for QR code reading to work.  I guess if you have a sparse QR code so the blocks are big you might have some chance...

[Damian Axford]

got OpenCV working,   now installing zbar library...    (brew install zbar)

then to rip-off something that already works...

[Damian Axford]

zbar install knackered on mac - can't be bothered to figure it out now

got the email capture working though...  might try making a timelapse tonight

given zbar issues, and QR code being overkill, going to have a look at basic fiducial markers in OpenCV using SURF descriptors.   there are some demo/example projects on tinternet that might provide a quick start.

[Jamie Osborne]

I would have recommended brew, but see you already use it. 🍻

However, when installing OpenCV, it wants to first install gcc as a dependency. I - and brew - normally compile with clang, and there is no hard dependency on gcc in the OpenCV formula!?

Did your brew install gcc along with OpenCV - and what does it usually use on your system (can't remember if clang ships as standard without any dev tools installed)?

Jamie

https://telegram.me/jmeosbn

On 17 Mar 2015, at 12:32, Damian Axford <dam...@axford.me.uk> wrote:

got OpenCV working,   now installing zbar library...    (brew install zbar)

then to rip-off something that already works...

--