Operation Close Pass

[Pete Owens]

We called them "Swedish lollipops" and used them a lot in the 80s and 90s.

I remember them. I never saw anybody other than utility riders using them, and very very few at that.

I've never used one, but if they 'add width' say 50cm, then (hopefully) another vehicle will give an extra 50cm?

But they didn't - they extended no further out than your right elbow.

To be effective they need to be much longer:
http://www.happycyclist.org/?p=1943

[millimole]

What happened to the old 'reflector on a stick' (sorry - I can't remember the correct name), as they always seemed a good idea.

Good question.

We called them "Swedish lollipops" and used them a lot in the 80s and 90s. I came across some in the garage at the weekend.

IIRC the swivel mounts broke and I never replaced them... certainly can't remember an active decision to stop.

Jonathan

My utility bike has one on it still. I never use it (it's tucked behind the pannier).
I remember doing a non-scientific study of my commute with & without it - there was no great difference in the passes, and if anything the lollipop made matters very slightly worse. This was explained to me as being due to the drivers focusing on the lollipop, not on making a safe pass. I think the term was 'target fixation'.

[alanesq]

I was planning to build a electronic distance sensor on the side of my bike to record the distance of passing cars but never actually got round to doing it. Would there be any interest/benefit in this do you think?
I thought that whilst never going to be admissible in court or anything like that, it might be useful if a few people had these in creating some interesting data on the issue?

[Jdsk]

I'd be interested: there isn't nearly enough research on this.

Previous work
Walker. What else? Let me know if you need help with the literature search.

Technology
Action camera
Parking ultrasound sensors
Loggers... do any have a channel into which you could feed the distance signal?
Arduino/ Raspberry Pi

Jonathan

[alanesq]

I bought myself a ultrasonic sensor a while back (jsn-sr04t), I have not tried it on the bike yet but seems to work pretty well and costs around £6 on eBay.
This is very easy to link to an Arduino (https://www.makerguides.com/jsn-sr04t-arduino-tutorial/) but I am thinking an esp32 camera module may be a good option as these are around £5 have built in sd card reader and would be able to capture a picture of the passing car as well as record the distance.
I will give it a try and report back but may have difficulty with it being 3.3v and very limited IO pins available...

Update: I now have the software working and next step is to attach it to my cycle and see if it looks like it will work. It continually reads distance and if this drops below a set threshold it takes a picture and stores it on to an sd card along with the measured distance.
BTW - Anyone interested I have posted my progress so far on github at https://github.com/alanesq/ESP32Cam-range

prototype v1.0

[cycle tramp]

What happened to the old 'reflector on a stick' (sorry - I can't remember the correct name), as they always seemed a good idea..

I still use mine - although its arguemented with a fairly up right position and a bar end mirror sticking out my right side handlebar end... I get motor traffic giving me more space since i fitted a bar end mirror.

[Jdsk]

It continually reads distance and if this drops below a set threshold it takes a picture and stores it on to an sd card along with the measured distance.

Could you change that to sample and record absolute distance frequently rather than using a threshold... that might provide additional information.

Jonathan

[alanesq]

It continually reads distance and if this drops below a set threshold it takes a picture and stores it on to an sd card along with the measured distance.

Could you change that to sample and record absolute distance frequently rather than using a threshold... that might provide additional information.
Jonathan

That should not be a problem, the beauty of building something like this is you can modify it any way you wish.
I plan to take it for a test ride tomorrow so depending if it looks like it has potential or not I will take it from there. The sensor works really well on my test bench but how it will respond to being on the side of a moving cycle remains to be seen...

[Jdsk]

: - )

Jonathan

[Pete Owens]

I thought that whilst never going to be admissible in court

I don't see why it shouldn't be.

[alanesq]

I gave it a try this morning. It shows promise as it did not just trigger all the time for no obvious reason which I thought it might.
I had it set to trigger below 1m and there seemed to be very little traffic when I was using it so no one passed anything like that close.
Ironically after I stopped using it a car gave me a punishment pass for daring not to be on the cycle path which would have been a great test for it - lol

I think my next test I will modify the code to just constantly record distance readings (as suggested) and will see if the data from this correlates with passing vehicles. It looks like it will be able to record a max of around 3 readings a second.
Correction: I realised there was a delay in my sketch which served no real purpose, removing this and I am now able to get something like 90 distance readings a second - that is a bit more impressive

[Jdsk]

Very interesting.

In the absence of a single system recording everything have you got a method for retrospective synchronisation of the data on distance with what a camera records? Something like a modern version of a clapper board?

Jonathan

[thirdcrank]

If this could be made reliable then IMO one really useful benefit of the resulting evidence would be to supplement camera footage to compensate for the effects of the usual wide-angle lenses which make everything seem further away.

There's no reason that this sort of evidence would not be admissible, but it would have to be totally reliable from the outset.

[alanesq]

I have just been out for a quick test on my bike. "luckily" I got reasonably close passed within the first 100 yards (at least you can rely on cars for something

The attached graph shows the captured distance data from this event. I am pretty impressed with this result, it shows that the cars distance was measured around 14 times as it passed and each reading agrees with the distance (around 1.2 meters).

btw - I have limited its range to around 2.5 meters so basically anything over 2.5m (or if no echo is received) it will just show as 5m, so you can see that one of the attempted readings as the car passed did not work (maybe reflected from a wing mirror??)

In this case it was taking a distance reading around every 33 milliseconds. So it took a total of 460ms or nearly half a second for the car to pass.
Interesting thought that you could calculate the cars relative speed from this. I just had a quick try at a rough estimate and came up with 28mph.

Anyway, I think this works as a proof of concept, is this of interest/use to anyone? If so I will carry on developing it a bit more otherwise I think I have done all I wanted to (I was just interested to see if it would work).

[Jdsk]

Neat.

Jonathan