back to article Network Rail steps back from geofencing over safety fears

Network Rail has become the latest company to flag the risks of smart technology implementations despite receiving a safety award for geofencing applications. Network Rail and Connected Places Catapult were given the award at an industry event in London last June. Three suppliers were part of the collaboration: OnWave, Track …

  1. Anonymous Coward
    Anonymous Coward


    ... working in dangerous areas is expensive. You need the worker (s) and then safety officers / supervisors who stay in contact with control / despatch and that adds an expense. Many have been searching for the holy grail of technology that will identify that someone is working on site, but that technology can't make judgements, only decisions based on inputs.

    Tag all the workers, have them chirp regularly, and how many missed chirps before you declare them as left the area, not in danger any more and reopen the line?

    It's a solution I and many others have been seeking for years.

    1. katrinab Silver badge

      Re: because...

      You would only reopen the line when all the workers have left. If for example 19 out of the 20 workers have left the area, and the other one has gone offline but is still in the area, that isn't good enough.

      1. Anonymous Coward Silver badge

        Re: because...

        But the gist of the article is that they want a system for use while the lines (tracks) are still in use. That needs to be far more resilient.

        Plus, I imagine any form of GPS based system won't work in tunnels and would have limited use inside buildings.

        1. Stuart Castle Silver badge

          Re: because...

          It also may not be that reliable in areas with lots of skyscrapers..

          1. Roland6 Silver badge

            Re: because...

            Interesting they tested it at Milton Keynes, perhaps taking a train down to Euston and walking it through Euston throat - no skyscrapers but deep cuttings and multiple levels of track, would be a good test.

        2. Roland6 Silver badge

          Re: because...

          > But the gist of the article is that they want a system for use while the lines (tracks) are still in use.

          Which requires location accuracy of 1ft or better and warning/alarm notification within seconds, combined with real-time update as the worker moves towards (or more importantly away from) safety.

          However lower level of accuracy can be useful, if used in conjunction with other railway safe working practises. From memory one of the big (and easy to make) safety mistakes workers make is to step the wrong way of a line with an approaching train, only to encounter a train travelling in the opposite direction…

          1. rg287 Silver badge

            Re: because...

            However lower level of accuracy can be useful, if used in conjunction with other railway safe working practises. From memory one of the big (and easy to make) safety mistakes workers make is to step the wrong way of a line with an approaching train, only to encounter a train travelling in the opposite direction…

            Indeed. There's a whole slew of possible working conditions from railways - from a total possession on the entire corridor (all lines closed) through to engineers working on the embankment with live traffic on all lines. And a host of intermediate states (some portion of lines closed, with or without a physical barrier between them and the live lines).

            Work is often at night (to minimise disruption), under harsh worklights which casts deep shadows, so workers might not see other workers waving furiously.

            Machinery is noisy, meaning shouts or radio calls might be missed.

            Geofencing is definitely an interesting and promising technology as part of a suite of tools to improve worksite safety. Just as they've adopted high vis, radios, cameras to cover machinery blind spots and other signalling equipment over the years, as well as having designated safety supervisors whose only job is to keep track of whether everyone else is following a safe method of work.

            There are no silver bullets though.

            1. Dave314159ggggdffsdds Silver badge

              Re: because...

              "There are no silver bullets though."

              There's the 500 Series Shinkansen...

              1. katrinab Silver badge

                Re: because...

                And the Frecciargento, which is admittedly a silver arrow rather than a silver bullet.

        3. Eclectic Man Silver badge

          Re: because...

          I use a Garmin Forerunner GPS tracker when out running. My regular route (down the road to the junction and back) is measured with about 200 meters variation between the shortest measurement and the longest one. I doubt that GPS with moving people will ever be sufficiently accurate for close supervision of safe location.

          1. Dave314159ggggdffsdds Silver badge

            Re: because...

            Your consumer device intended to give approximate tracking isn't comparable to the higher-precision devices that exist. (Pedantically, they're all accurate, some are more precise than others.)


            Dual-frequency GPS devices can reliably resolve to within a few cm on-the-fly.

            1. Eclectic Man Silver badge

              Re: because...

              The most accurate, affordable GPS unit I could find is the Garmin GPSMAP 66st which is reputedly accurate to a few metres*, but can be 10 to 20 metres out for a few minutes (my emphasis).

              "I have noted on other garmin GPS units, but not yet on the GPSMAP 66st, that sometimes (perhaps 1/5 of the times) the unit shows a good accuracy (3-6 m), but the pointer and track are a bit off the actual location. It is often around 10-20 meters. After 5 to 20 minutes, it suddenly corrects the position to be much more aligned with the actual shown accuracy. It is especially easy to see if one takes two or three laps on a route/track which is about 15 minutes long per round. Running a loop, driving back and forth a few times, etc.

              I read that one should allow the unit to get its signal for 30 minutes to make sure accuracy is good. This seems to be a good general rule of thumb.

              The land survey authority has points all over the country (sweden) and i have one not too far away which i checked against, its position is given to an accuracy down to 0.15 centimetres.

              I marked one point and a second which was averaged for 5 minutes. GPSMAP 66st (GPS+GALILEO and 3 satellites reported EGNOS, D) and a etrex vista hcx (GPS, but no EGNOS reported) was was used.GPSMAP 66st, showed 2.4 m accuracy. The etrex 3-4 m accuracy.

              The differences between the non-averaged point and the averaged is negligible on both devices.

              The reported location compared with the real location is 4-5 meters on both devices. Closer to 4 m on the GPSMAP, and closer to 5 m on the Etrex."

              Garmin state:

              "Garmin® GPS receivers are accurate to within 15 meters (49 feet) 95% of the time. Generally, users will see accuracy within 5 to 10 meters (16 to 33 feet) under normal conditions."**

              Is that good enough for the railways? Professional, ie Ordnance Survey and other map maker quality, GPS sets cost thousands. I suspect that a few minutes of being 15 to 20m away form your recorded location may be unacceptable.

              *,5%20meters%20on%20both%20devices. (Post by '987' Sun Dec 20, 2020 1:38 pm)


              1. Dave314159ggggdffsdds Silver badge

                Re: because...

                Did you read the link I provided? It answers all your questions.

                The short version is that dual-frequency devices are more expensive than consumer devices usually are, but that isn't a big deal in terms of rail infrastructure spending. Other methods of GPS augmentation are cheaper and may be enough under the specific circumstances.

                FWIW, there is no significant difference in the precision of single-frequency GPS receivers. Well, maybe some shonky ones are even less precise than most. But mainly it's a physical limit, and not something that changes significantly until you get to the point of using dual-frequency receivers.

                1. Eclectic Man Silver badge

                  Re: because...

                  Yes, I did read the link you provided. Did you read the link to the Garmin site regarding their accuracy that I provided?

                  The fact is that when mapping organisations such as the UK's Ordnance Survey use GPS to check a location they leave their GPS equipment there for several minutes to get an accurate fix (to within a few cm initially). On a railway work site where there is a possession (railway term for controlling a part of the track and preventing use by trains) of the track it is not static workers who are in danger but ones who wonder off. It can take several minutes for even a three frequency GPS device such as the Garmin devices on the link to catch up with the change of location. They can be out by 10 to 15 metres for several minutes, which is very dangerous on a line that is in use. That distance can be the entire width of a cutting or half way across a marshalling yard. The point about using GPS for precise geolocation on a railway line is that it needs to be accurate to about one metre for safety all the time, not just when someone has been standing still for 10 minutes.

                  1. Dave314159ggggdffsdds Silver badge

                    Re: because...

                    If you read it, your reading comprehension is evidently very poor, because it covers these exact points.

                    Obviously what you're saying is completely untrue; milspec GPS needs high precision during movement, and as the US gov site explains, this is entirely possible, but too expensive for consumer level devices. Static devices can achieve even greater precision.

                    It's telling that you still don't acknowledge the difference between precision and accuracy.

                    1. Eclectic Man Silver badge

                      Re: because...

                      "Static devices can achieve even greater precision." At least we agree on that. The problem with managing a perimeter on an engineering site is that the people move about, so are not static

                      From the link Dave314159ggggdffsdds provided:

                      "How accurate is GPS?

                      It depends. GPS satellites broadcast their signals in space with a certain accuracy, but what you receive depends on additional factors, including satellite geometry, signal blockage, atmospheric conditions, and receiver design features/quality."


                      "Why does GPS sometimes show me in the wrong place?

                      Many things can degrade GPS positioning accuracy. Common causes include:


                      Satellite signal blockage due to buildings, bridges, trees, etc.

                      Indoor or underground use

                      Signals reflected off buildings or walls ("multipath")"


                      "What is the government's commitment to GPS accuracy?

                      The government is committed to providing GPS at the accuracy levels specified in the GPS Standard Positioning Service (SPS) Performance Standard.

                      The accuracy commitments do not apply to GPS devices, but rather to the signals transmitted in space. For example, the government commits to broadcasting the GPS signal in space with a daily global average user range error (URE) of ≤2.0 m (6.6 ft.), with 95% probability, across all healthy satellites in constellation slots. Actual performance is typically much better. On April 20, 2021, the global average URE across all satellites was ≤0.643 m (2.1 ft.), 95% of the time."

                      So the target for the signal is to be accurate to within 2m 95% of the time.

                      As I said the best hand held GPS receiver I could find was the Garmin, yes there are more accurate ones, for example the Juniper Systems Geode GNS3*, but that is not exactly a user-friendly device for someone who is not a surveyor -. the images show it mounted on a pole, and it is hard to believe that railway workers would carry such a thing around while doing their engineering work.

                      I suspect we are arguing at cross purposes. You claim that there are highly accurate GPS receivers (which I accept there are), I am saying that conditions on railway lines and engineering sites and the usability of the highly accurate receivers makes them unrealistic for tracking people in real time with sufficient accuracy for safety. If you disagree, please provide a link to a highly accurate GPS receiver that can be easily carried an used by someone who is doing work other than surveying, in a built up area (such as a railway cutting or track between high rise buildings) where signals may be reflected or blocked.


                      and (see video)

                      1. Dave314159ggggdffsdds Silver badge

                        Re: because...

                        I think you're right about cross purposes. Let me try to break it down.

                        First, gps precision is possible, and cost isn't really a barrier here.

                        Second, GPS augmentation is the way to go.


                        It is not hard to conceive of systems that would add to a GPS location in useful ways in this context.

    2. Mike 137 Silver badge

      Re: because...

      I remember a conference presentation a few years back of a cunning smart card based system for ensuring that tower crane operators could only operate cranes within parameters defined by their level of training and certification. The idea was that the operator's card, once inserted into the crane's console, would limit the extent of its operation. At question time, I asked whether there was any mechanism for ensuring that someone could not use another (potentially better qualified) operator's card. Silence ensued, then the speaker suggested that this was down to site wardens to check when workers reported on site. Incomplete solutions like this abound in the race to automate.

      1. Dave314159ggggdffsdds Silver badge

        Re: because...

        Safety rules not working as well when people deliberately circumvent them isn't new. It's a related problem; ideal safety systems can't be circumvented.

        It's analogous to machine tools which could take a hand off: a rule requiring workers to put both hands on a bar clear of the machinery isn't as good as a machine that takes two hands to trigger.

      2. Jason Bloomberg Silver badge

        Re: because...

        a cunning smart card based system for ensuring that tower crane operators could only operate cranes within parameters defined by their level of training and certification.

        That's an inherent problem of automated enforcement; there's always some means of defeating it if determined enough.

        When it comes to providing 'additional protections' to what already exists it's a different matter.

        It's the difference between protecting others and protecting people from themselves.

        1. matjaggard

          Re: because...

          > there's always some means of defeating it if determined enough

          Well yes but you actually don't need to make it really hard to circumvent a protection, just harder than the alternatives.

  2. Sam not the Viking Silver badge

    Safety in dangerous areas very difficult.

    Safe systems of work deserve time and experience being spent on them. In my view, on-site supervision by experienced personnel is essential. I can't really see technology doing the supervising but I can see robots doing the actual work.

    Whilst revisiting safety procedures is important, it is more important to keep the object in mind.

  3. Anonymous Coward
    Anonymous Coward

    Incoming in 3...2...1...0

    The latest AI solution for the job of the 'Banksman'.

    Most of the time they are the gatekeeper of PW works. Their experience is paramount to keeping the workers safe. I'm sure that someone will invent an AI model that will do away with this otherwise unproductive member of the PW crew.

    IMHO and experience of working on the PW on a preserved railway, that would be total madness.

    PW - Permanent Way

    1. Dave314159ggggdffsdds Silver badge

      Re: Incoming in 3...2...1...0

      It isn't a replacement, it's an augmentation to improve the performance of the banksman.

    2. Mike Pellatt

      Re: Incoming in 3...2...1...0

      I've yet to see "banksman" as a label of the function of anyone working lineside in the (depressingly numerous) RAIB (Rail Accident Investigation Branch) reports into trackside worker injuries and deaths.

      1. FunctionalSaferyEngineer

        Re: Incoming in 3...2...1...0

        That's because banksman is a term not used lineside (and isn't adequately defined anywhere for off track civils work) the same job is performed by a Machine Controller MC or Crane Controller CC for lifting operations. These are terms you will find in the RAIB reports.

  4. Missing Semicolon Silver badge

    Why not alarm the dangerous thing?

    Put a leaky feeder along the track whilst work is in progress. Any one in range of the feeder will get ponged as being in danger.

    1. Anonymous Coward
      Anonymous Coward

      Re: Why not alarm the dangerous thing?

      What about cattle prods?

      Asking for a BOFH...

  5. xyz Silver badge

    These things are not safety certified...

    Except in the salesperson's imagination. Management should know this.

  6. Korev Silver badge

    AI solution?

    Surely all they need to do is train an AI solution...

    1. David 132 Silver badge

      Re: AI solution?

      Blockchain. That's the answer.

      Oh sorry, is it 2024 now? I struggle to keep up.

      1. matjaggard

        Re: AI solution?

        Ooh that's annoyed me, well done.

        For what it's worth I'm going to rise to the bait.

        Not all these technologies are all hype and no substance. Blockchain always had pretty limited use cases in my opinion. AI on the other hand is going to be pretty transformative to the world - equivalent to the industrial revolution or the internet. People will get crazy rich from it and very many, possibly most current jobs will just stop and we'll almost all have to change how we work to augment the AI rather than instead of it.

  7. andy the pessimist

    uwb could do it

    There is an uwb device inside an apple phone. With anchors in the right places it can locate your position precisely. If anyone has done software for this I don't know.

    I just test the devices.

  8. Norfolk N Chance

    Beeching's revenge

    GPS is just simply the wrong solution.

    It's amazing the accuracy which can be achieved under ideal conditions, within a few millimetres (it's revolutionised surveying). This does however require secondary static base stations (easy) and the primary device (wearable?) to have a decent low horizon view. Build it in a hat? Not much good when you have to bend over to work.

    Physical barriers to prevent access to dangerous areas ie live lines are the only reliable way.

    Sadly I don't suppose the humble (physical) fence has the wow factor these days.

    On that note I'm surprised someone hasn't suggested train-mounted LiDAR - a fantastic idea if it could alert the driver of an obstacle 2.5 miles away around a bend...

    Maybe they could use AI to defend them from the manslaughter charges or Bitcoin to pay the fines and compensation?

    1. Roland6 Silver badge

      Re: Beeching's revenge

      > a fantastic idea if it could alert the driver of an obstacle 2.5 miles away around a bend...

      320 kmh to 0 in 38 seconds…

      “ Emergency brake test on a TGV: 3389 metres to stop from 320 kmh!”

      3389m = 2.11 miles

      However, for maximum safety, it would need clairvoyant powers so as to apply the brakes 2.5 miles before anyone stepped into its path.

      1. This post has been deleted by its author

      2. This post has been deleted by its author

      3. Roland6 Silver badge

        Re: Beeching's revenge

        A quick check indicates the video either doesn’t cover the entire stopping time or the stopping distance is misstated.

        Whilst TGV clearly comes to a halt after 38s of video, a quick calculation shows that a train travelling at 320 kmh would cover 3378m in 38s.

        Counting gantries (26) and assuming typical distance between is 50(*), gives a total distance of circa 1300m in 38s.

        (*) typical spacing of overhead line gantries for 140mph(225kmh) in the UK are 64~73m. Eurostar (300kmh) it’s circa 50m.

        Some additional searching provides a calculated stopping time of under 90s, based on a number of assumptions, which seem more reasonable.(in the event of total power failure it might be slightly faster).

        This thus gives a distance covered of 8000m (5 miles) at 320kmh. Thus that 2.5 miles forward visibility is an understatement..

      4. FunctionalSaferyEngineer

        Re: Beeching's revenge

        It would also require a signalling system such as ETCS to only work 2.5 miles ahead. With the still relatively Victorian system in the UK it would need to set the protecting signal to red and they don't come around all that often, in some parts of Wales and Scotland there can be tens of miles between signals.

  9. teebie

    "ALO (All Lines Open) warning devices."

    "ALO (Any line Open)."

    Why does ALO have 2 different meanings.

    Is it me, or is the first half of the article hard to follow?

    1. FunctionalSaferyEngineer

      It doesn't ALO is Any Line Open, it was previously Adjacent Line Open but has NEVER been All Lines Open. That seems to have been a mistake by the author of the article.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like