EDS for Zoned ISB Routes: In Pictures

Reprogramming the electronic destination signage on our Volvo B9Ls is not easy. The display units are connected to the GPS and next-stop announcement system, and built by a Shenzhen company called KABYM which doesn’t seem to provide its customers with an easy-to-use software to program their own data.

So when COVID-19 hit, and we had to re-draw the entire ISB network, at NUS we all more or less gave up and used a bunch of laminated A3 paper sheets to cope with the first few weeks, while the EDS displayed “COMFORTDELGRO BUS” all the way.

For routes CC, CCX and DD1/DD2, the destination of the bus was displayed above the route letters in order to disambiguate the direction. For example, CC buses had either “UNIVERSITY TOWN” or “OPP KR MRT STATION”, which if you ask me is way too many letters to be legible.

First dataSet: Scrolling Words (All Buses)

In end-July, new EDS controllers were fitted to allow the ISBs to display the zoned routes (AA1, AA2, BB, CC, CCX, DD1 and DD2). The original remote-control-style controller, known as the Kabym NS-K, was replaced by the Kabym AL-K80, a large touchscreen tablet that was easier for the bus drivers to use.

It appears that whoever did the reprogramming took a less sophisticated approach. The font used was no longer Arial, as used from 2015-2020, but a serif font that I presume is a default Chinese font (it resembles SimSun). The ability to use GPS location was not taken advantage of by this programmer, so there was no automatic switching of the display mid-route or displaying of “Please do not board”.

The new EDS data suffered from quite a few bugs. This is entirely understandable to me, and I don’t blame the programmers one bit for the issues. It is likely that due to the rush, some miscommunications occurred or some details were overlooked, and I don’t want to blame anyone for this. Nevertheless, I shall list some of them here anyway seeing as they were fixed swiftly.

One such issue was that instead of being wrapped into 2 lines to fit inside the screen, the route details scrolled across the entire EDS moving slowly. This made them rather hard to read, as one could not take in the route of the bus at one glance. Instead, they would have to stare at the EDS for about ten seconds to read the entire string. For example, in the photo below, the words have just started coming in from the right and move slowly leftwards.

Then, some services had problems with the route details themselves. For service CC in particular, which doubles up at COM2, there is a need to distinguish whether the bus is heading from Opposite KR MRT to UTown or the other way. Initially, the goal of distinguishing the direction of travel at was not achieved by the EDS. Instead, the entire route starting and ending at Opp KR MRT was reflected on the moving string, which showed “Opp KR MRT > COM2 > UTown > BIZ2 > PGP > Opp KR MRT”.

As a result, drivers using the CC EDS had to continue using the paper signage particularly when calling at COM2, or else take questions from passengers. I suspect that the programmers engaged by ComfortDelGro were simply unaware at the time that CC had this specific need.

There were also some typos or data that did not make sense. Service BB buses displayed a correct string, but it left out UTown, showing the bus as plying “Opp UHC > UHC”. Service CCX had a typo in the details where “Opp KR MRT” was mistyped as “Opp FR MRT”.

The interior next-stop display was inaccurate, often jumping straight to displaying the final stop after only a few stops had been visited. Many buses would display “Next Stop: PGP Residences” after just a few stops. I suspect some of the GPS waypoints for bus stops on two different sides of the road had been placed too near each other, causing the system to think that the bus was already on its way back.

As a result of these bugs, many drivers were advised not to activate the signage, while waiting for ComfortDelGro staff to assist in reprogramming the system. Exceptions were often made for nighttime, however, as an inaccurate EDS was better than no EDS in low-light conditions to increase visibility for passengers (the paper signage was difficult to see), and some drivers ended up using both the preliminary EDS and the paper signage (which benefited students running for the bus from behind).

Second dataSet: wrapped words, fixed cc (PC3906L, PA33K)

In early August, PC3906L was borrowed for testing as part of an effort to reprogram the signage. For three days, it ran around NUS with a “testing” signage placed in front, in an effort to properly map out the bus stops and fix the problems.

From this point on, it became a “guinea pig” for testing new ISB data. It returned to service on 16 August, having had the EDS data fully updated. Under the zoned ISB route arrangements, PC3906L is actually rostered as one of the few buses that does not stick to one route for the whole day, instead performing trips on every service. I wonder if this was why it was selected to be the first bus to receive the new data.

The new data fixed quite a few issues. The route details for each route have been simplified so they do not report the entire list of bus stops, and they also spill into 2 lines and stay on-screen without scrolling.

For CC, the display was simplified. It now actually differentiated between “Opp KR MRT > COM2 > FASS > UTown” and “UTown > FASS > COM2 > Opp KR MRT”. This was still not ideal to me as it used up two lines of small font. It was not automatically toggled by GPS. However, it is already much better than the original CC operation.

Sometime later, PA33K followed suit. PA33K was relegated to a spare bus under the zoned deployments, so I only observed it on 4 September 2020 when it was used to replace PC3989Y on AA2.

These updates were not immediately pushed out to all buses, so the second dataset was seen only on PC3906L and PA33K.

Third Dataset: Auto-toggle (PC3906L, PA33K)

Over the weekend of 12-13 September, an attempt was made to update the signage for CC such that it did not need to be manually toggled at UTown. I believe this update also changed the font to a slightly smoother one without serifs. This update was first pushed to both PC3906L and PA33K.

However, it soon became apparent (at least to me) that there was an issue. When I observed PC3906L doing its CC trips on 14 and 15 September, I noticed that the bus departed Opp KR MRT station with the correct EDS showing the first half of the route (Opp KR MRT > COM2 > FASS > UTown).

However, when it entered COM2 for the first time, the display actually showed a two-line version of the entire route. That was the same string (Opp KR MRT > COM2 > UTown > BIZ2 > PGP > Opp KR MRT) that would move across the screen in the very first programming.

When PC3906L was at UTown, or at COM2 on the second visit, it showed the correct string for the second half (UTown > FASS > COM2 > Opp KR MRT), though I think I remember seeing it show the whole-route string when it was at BIZ2.

I gave OCA and CDG a heads-up, just in time to run into PA33K. It had the same updated CC EDS, and I nearly boarded it until I saw the “Testing” paper and the technician accompanying the bus captain. This meant that thankfully, ComfortDelGro and the technicians were aware of the issues and were testing the display to rectify it.

I’m not sure if the intention of the programmers was to perhaps display the entire route most of the time and switch to only the relevant half whenever the bus entered COM2, but I think that might have been a bit too much complexity for the EDS system to handle.

Third Dataset: Rollout (All Buses)

During Recess Week (starting 21 September), the finalised EDS programming was now rolled out to all buses. I think ComfortDelGro chose to update most of the buses first, so I saw AA1, AA2, BB, DD1 and DD2 all using the EDS while CC buses continued to use paper signage for most of Monday (21 September). The final bus to have its EDS updated was PC3989Y, which ran on AA1 with the old format on Monday morning and had the EDS updated in the afternoon.

Compared to the second dataset, DD1 and DD2 now had centred route details.

Now that the EDS data rollout was widespread, I was able to observe that between Opp KR MRT and PGPR, the EDS displays “Please Do Not Board” on the front and side, instead of route details. I believe this feature was programmed in in the third dataset. Previously, EDS for routes A2 and D2 had this functionality.

Occasionally I did see one or two drivers press out the EDS for CC (although they made sure to use the paper as well), and it had a hilarious amount of mojibake on it. In the photo below, the driver removed his CC paper as he was at PGP, going to end his trip at Opp KR MRT, and he changed the display to “Off Service” immediately after leaving PGP. It made for quite a funny sight.

Thankfully, on Tuesday (22 September), CC buses finally had their controllers updated with working CC EDS data, which simply showed the first half of the route during the first leg and automatically switched to the second half of the route during the second leg.

More information on the entire evolution of ISB route signage can be found at this article.