Auto means auto
Conflict of interest alert: I may or may not want to get a job building robots.
I apologize for the lack of content, but with the vise of increasing virus-related measures clamping down I’ve been trying to get as much work out of the way as possible. Until now, that is, since with the totally-not-a-lockdown I’ve got some time to think.
This may not make me a lot of friends in the relevant halls of power, but I’ve got to say it anyway. Vancouver. Copenhagen. Paris. Even KL. All these cities have achieved fully automated, unmanned train operations.
So why can’t we? Why do I see a fellow in SMRT uniform leaning against a wall at my local LRT station, as well as aunties pointing and calling on our automated trains?
I did mention previously that Singapore was arguably one of the leaders in the world in terms of automated railway operations. But in fact, in recent years, an interesting regression has occurred. Absurd as it may sound, we are putting people on trains that are supposed to be able to drive themselves.
Let’s be frank with ourselves. For all our sakes, this reckless consumption of manpower has got to stop. I did briefly explore previously what “auto means auto” actually meant, but better late than never to take a closer look into the entire issue, especially now that I have a better idea of what capabilities and concerns we actually have, and given the current climate of surrounding matters.
“Look East Policy”
I like to call the good doctor and friends “government by weeaboo”, but maybe it may be a better idea to look at my own country first. People in glass houses shouldn’t throw stones.
In Japan, the classic solution to a problem is to throw as many warm bodies at it as possible. For what it is worth, though, they’re starting to find out that this is unsustainable. It then shouldn’t come as a surprise that they’re trying to replace frontline staff with robots.
We also see similar things in Singapore with unmanned convenience stores. There’s one in NUS Frontier canteen that I’ve seen personally. So how do these convenience stores deal with crime? Incidents such as someone setting the microwave on fire? How about a power outage? This is probably on a different risk level as something like public transport that’s nationally critical, but answering those questions is worth a start.
I personally feel that this has become a lot more important in light of the Wuhan virus and Malaysia’s lockdown forcing Malaysians working here to hastily scramble for a place to stay, when shuttling across the border to their homes in JB was no longer an option with the closed border. Much was made about bus drivers getting to stay in five star hotels, which were already buckling under the shutdown of tourism already. But it begets asking, is this sustainable, like what the Japanese are learning?
Heck, even the Taiwanese (yes, KBW, you listening now?) threw a ruckus when it was found out that Taipei Metro was spending NT$140 million on hiring staff to man their automatic trains on the Wenhu line. That was in 2010; I don’t think they’re still doing that today.
Running before you can walk
To begin with, the government has already been pushing quite heavily for autonomous vehicles. At ITS Congress 2019 last year, I had the opportunity to ride on various autonomous vehicles, like the ST Engineering and NTU self-driving buses. Of course, there are also the self-driving bus trials going on at Sentosa and at NUS, or self-driving cars in the one-north area. One thing struck me, though — they all have safety drivers, supposedly a requirement under the law.
The reasoning is clear — the roads are highly unpredictable environments, with rash drivers, jaywalkers, and whatever other dangers there are out there. Logically speaking, having a safety driver to quickly take the reins when there’s something the computer can’t do, makes sense.
Continuing on, it stands to reason that the railway, especially a fully underground subway line where weather is not a concern, should also be a far more controlled and controllable environment than the road. Furthermore, unlike in Nuremberg and Vancouver, in Singapore all stations are equipped with platform doors that prevent intrusion onto the tracks.
Thus, logically speaking, you would want to master the unmanned railway first, before applying some of its lessons onto unmanned buses and other road vehicles. After all, it stands to reason that the technical and social challenges of the unmanned railway would be similar to the challenges of an unmanned bus or car on city streets.
Furthermore, the Committee of Supply (how long ago did that seem?) was already acting like bus drivers were going to be put out of work within the financial year. Well, we’re not there yet, though I can understand if it’s a political move meant to assuage concerns from the union.
Walking before you can run
But what are they? I’m thinking things like passenger safety and security issues. Or quick recovery from failures so they don’t go past the 5 minute mark and tank the all-important MKBF.
Highly ambitious targets have to be met by highly ambitious strategies, or perhaps the other way may have to be to curb one’s enthusiasm. In my opinion, the issue here is the same issue as why people queue for days on end to buy iPhones and willingly cough up the Apple tax — statistics. We may want to find a nice compromise between meeting highly ambitious statements to get politicians reelected (do they even have to?), and cost effectiveness.
I meant to actually go on a ”study trip” to Vancouver on my way to exchange in America and see just what they’re doing there. After all, the Vancouver Skytrain has been operating fully-automated even before our first MRT line in 1987. But obviously the Wuhan virus put paid to that. Thanks, Pooh! In any case, we can still get somewhere with data available on the internet. Staying at home also means we can do a fairer comparison with the Copenhagen Metro.
The first thing to note is that Vancouver achieves only a 97% on time performance. Copenhagen reaches 99.2%. In Singapore, our Circle Line reaches a round 100.0%. Of course, it bears noting that Vancouver has much more lenient standards than we do, since a train is only considered late there if it’s late by more than 3 minutes, compared to 2 minutes here.
I don’t particularly like this next statistic because I don’t think it really means anything, but I’ll have to discuss it anyway. Copenhagen only achieves a MKBF of approximately 20,000, which is far below the million mark. As for Vancouver, they managed to hit half a million in 2018, though they’ve changed the metrics in 2019 and that doesn’t look very good at only about 250k. It’s important to note that measurement methods may differ, that’s why I don’t like this number.
Of course, an argument can be made that density here and high reliance on public transport means we must demand a level of reliability much higher than Vancouver or Copenhagen. But are you prepared to pay for that?
The old saying goes, shoot for the moon and you end up in the clouds. Highly ambitious targets may help to rally the troops for a common goal. But is it really necessary in the long term, or can it do more harm than good?
There’s an elephant in the room, and he’s lime green
The above arguments of course work on most of our MRT lines, which tend to have short distances between stations, and thus ample opportunities for intervention when something goes wrong.
But what about the CRL? CRL will see long stretches of plain track with nothing in between. Such as around Changi Airport, Paya Lebar Airbase, and of course the CCNR. Furthermore, CRL will operate 6-car trains, with headroom to go to 8-cars in the long term. This makes it difficult for outside interventions to take place, although the latter issue is probably already a thing to iron out.
Now, currently, the train driver on NSEWL trains is able to access all the necessary information on the train and the passenger cabin, such as diagnostic information and CCTV feeds, from the comfort of his cab. On automated lines, though, all this functionality has to be packed into the small driving console, which is also not accessible to the manned staff on the trains at all times (since the public can access it, right?)
It makes things interesting for a staff on the automated train. How is he able to quickly access information to resolve a situation? Fight his way through a crowded peak hour train car in order to get to the driving console? Or worse, in a security situation, how can he be sure that he is in a relatively safe position? Or how does he resolve an issue with car 5 when he’s in car 1 and without easy access to the PA system to speak with commuters?
Consequently, it means that the LTA has a few hard decisions to make before CRL opens in 2029; in fact, they may have to make it soon, since the CRL systems should be under design at this point in time. Do they provide the CRL trains with a driver, or improve the automated systems such that the automated systems are able to do the heavy lifting work that the driver does?
Or, to go the full nine yards, provide to the staff augmented reality devices that can connect wirelessly to the train systems? But maybe that’d be another can of worms.
I leave you with this: How robot are our robot trains now? How robotic should they be?
