Slashdot Mirror
The Fourth US Navy Collision of the Year Was Ultimately Caused By UI Confusion (arstechnica.com)
Yesterday, the U.S. Navy issued its report on the collisions of the USS Fitzgerald and USS John S. McCain this summer, which was the fourth U.S. Navy collision this year. "The Navy's investigation found that both collisions were avoidable accidents," reports Ars Technica. "And in the case of the USS McCain, the accident was in part caused by an error made in switching which control console on the ship's bridge had steering control. While the report lays the blame on training, the user interface for the bridge's central navigation control systems certainly played a role." From the report: According to the report, at 5:19am local time, the commanding officer of the McCain, Commander Alfredo J. Sanchez, "noticed the Helmsman (the watchstander steering the ship) having difficulty maintaining course while also adjusting the throttles for speed control." Sanchez ordered the watch team to split the responsibilities for steering and speed control, shifting control of the throttle to another watchstander's station -- the lee helm, immediately to the right (starboard) of the Helmsman's position at the Ship's Control Console. While the Ship's Control Console has a wheel for manual steering, both steering and throttle can be controlled with trackballs, with the adjustments showing up on the screens for each station. However, instead of switching just throttle control to the Lee Helm station, the Helmsman accidentally switched all control to the Lee Helm station. When that happened, the ship's rudder automatically moved to its default position (amidships, or on center line of the ship). The helmsman had been steering slightly to the right to keep the ship on course in the currents of the Singapore Strait, but the adjustment meant the ship started drifting off course.
At this point, everyone on the bridge thought there had been a loss of steering. In the commotion that ensued, the commanding officer and bridge crew lost track of what was going on around them. Sanchez ordered the engines slowed, but the lee helmsman only slowed the port (left) throttle, because the throttle controls on-screen were not "ganged" (linked) at the time as the result of the switch-over of control. The ship continued to turn uncontrolled to port -- putting the ship on a collision course with the Liberian-flagged chemical carrier Alnic MC.
At this point, everyone on the bridge thought there had been a loss of steering. In the commotion that ensued, the commanding officer and bridge crew lost track of what was going on around them. Sanchez ordered the engines slowed, but the lee helmsman only slowed the port (left) throttle, because the throttle controls on-screen were not "ganged" (linked) at the time as the result of the switch-over of control. The ship continued to turn uncontrolled to port -- putting the ship on a collision course with the Liberian-flagged chemical carrier Alnic MC.
See? This is what happens when the project team is made up of "full-stack" developers - no one knows how to code a decent UI... That, and the whole thing was written in Angular...
If you want news from today, you have to come back tomorrow.
Yes, the displays could be improved, but there were many issues involved. The proper crew members were not on the bridge. The captain had let them sleep an extra hour, so the most experienced crew were still at breakfast. The crew involved were not experienced enough to make quick changes to the ship configuration in a critical situation. Lesson learned, don't get in a critical situation you know will happen with the back up crew. Another good point - don't have just one set of high quality bridge crew-members.
https://www.youtube.com/watch?...
Who's controlling it?
What's "it"?
The ship!
You are, right?
Only the rudder.
Whose controlling the propellers then?
I thought you were.
No, I thought you were!
Let's ask Mikey. Mikey, are you controlling the propellers?
I was earlier, but I thought you guys took control of them. See, the red icon is on.
That's not a red icon, that's the object we are about to collide with...
Table-ized A.I.
Just design, humans, contractors and internal gov/mil policy.
Who would have expected not having a good design and testing that design with crews would have been an issue?
How to do a "navy".
1. Your crews have to have skills. Find the best people to work in your navy. Give them the wages, support and education they need.
2. Read up on how other winning nations did the "navy" design over the many, many years.
Saying a ship is new or a different "design" is no excuse.
The UK built its Dreadnought https://en.wikipedia.org/wiki/... having to consider new designs and new ideas. The new parts got made with skill and people worked very hard to get the new design ready.
Any unexpected issues got corrected by real engineers and top experts before they became an issue for the navy.
Find the experts, test things a lot, have good crews and ensure the skill sets are ready.
Domestic spying is now "Benign Information Gathering"
Despite Ars Technica's single-minded view on the incident, there were multiple levels of failure, including but not limited to:
Insufficient lookouts
Overcrowded bridge interior
Insufficient training(What Ars Technica neglects(possibly deliberately) to mention is that parts of the crew on watch on the bridge were on temporary assignment from the cruiser Antietam, which according to the report has a different control system)
Contrary to protocol, the CO issued orders, without announcing that he was taking direct control, and then didn't keep a firm grasp of the situation.
Insufficient bridge watchstanding experience in general in the US Navy officer corps, partially due to the generalist nature of US Navy surface officers, rather than the specialization found in the Royal Navy for example. As highlighted by a USN officer: "âoePeople squeak through the system. They may be great officers and they may great engineers, but they might not have had a lot of time handling ships in busy waterways,â McGuffie told USNI News in an interview.
âoeWe have guys that are commanding ships right now that have 400, 500 hours of bridge watchkeeping time in their career.â
In contrast, as the bridge officer on a Royal Navy frigate for a six-month deployment, McGuffie stood watch for more than 2,000 hours â" all of them logged"
The Fitzgerald was both better and worse. The OOD had 0 situational awareness, ignored technical tools such as AIS that'd have given him sufficient situational awareness, ignored warnings by the junior OOD, insufficient lookouts posted(none on starboard side), the OOD had no knowledge of the TSS(despite being based out of Yokohama!!!!), and the TSS was not mentioned in the navigation briefing.
And, as on all USN surface warfare ships, non-pilots seem to be chronically sleep-deprived.
So, you have systematic issues at multiple levels, of which the UI was just one small part
WTF is this, star trek?
They don't need reconfigurable controls, do they? Wouldn't it make more sense to have discrete controls? They can be electronic displays if that's your thing, but probably shouldn't be.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
> The helmsman sent all control (not just throttle) the the other station.
And neither he nor the anyone else looking at the situation didn't realized all control had been sent away, because the UI didn't gray out of the inactive controls or anything. Two people looked at it and couldn't tell it had been inactivated. Guess which controls are disabled here:
https://upload.wikimedia.org/w...
> The second helmsman throttled down only one engine.
When he too couldn't tell that a) he had control of steering or that b) the engined weren't ganged. Again, try to figure out which controls are ganged and which aren't:
http://3.bp.blogspot.com/_IOWi...
It's not hard to make it obvious.