5 ms·

Remind me when USS McCain collided with another ship in a busy channel [0]. Because of a UI issue, they couldn't figure out where throttle control was and ran i

• kayodelycaon 5y ago
  I think there are several orders of magnitude more planes than submarines. From Wikipedia: Los Angeles class submarines built: 62, entering service 1976. Abus A350s built: 443, entering service 2015.
  • tra3 5y ago
    Fair enough, I suppose. The article author was comparing submarine nav software to phone nav software though and google says there are 5+ billion mobile users. That's another couple of order of magnitude more. That'd explain why mobile software is more polished. I guess any kind of mass market product would be higher quality. The same argument ("number of eyeballs") has been used with open source software vs closed source.
• jwithington 5y ago
  There are probably a few thousand users of submarine navigation software. But the willingness to pay *should* be quite high, which in most other markets would translate to a quality product. The tool should help submarines: * Prevent groundings (loss of a $3B platform, 120+ lives in worst case) * Do their missions better (increase the value of the platform)
  • tra3 5y ago
    They probably have a laundry list of requirements (like nuclear containment?) before they even get to software. I've been in many a design meeting where a "must have" was pushed out to the next version because "the user can just do this manually for now".
• hguant 5y ago
  The UI issue is real, and is a problem, but the root cause of the 7th Fleet's crashes during that time period is lack of training and inadequate leadership, due to an operational tempo that's Thenot sustainable given current funding and ships. Instead of building new cruisers, or taking the time to dry dock and repair these ships, the admiralty has been ok'ing "at sea" repairs, leading to situations like this where primary systems aren't functional and back up systems are run for far longer than intended. The admiralty has been looking for technical solutions to a human problem for about a decade now, becaise they'd rather buy a new set of carriers and the F-35 than build the ships they need to support the mission given to them by Congress. The result is burnout, and incidents like this. The Navy as a whole has a severe manpower shortage, and in the incident report for the McCain collision I believe it came out that the officer on duty was 17 hours into their watch, and didn't have a full compliment in the watch house because one of the crew who was supposed to be there was off in the lower decks for an unstated purpose (anecdotal evidence from my Navy friends, probably catching up on sleep, or seconded to repair something while they had "non-critical" time). The collision only happened because instead of being trained on seamanship as a whole, incoming officers are trained how to work the computer systems they're interfacing with. The Navy cut down the training periods as well - instead of training under a senior officer for a few months, to offset manpower shortages, incoming officers are put in charge immediately, and given a stack of CDs (wish I was kidding) to complete their training. The surface Navy is in an incredibly dysfunctional place right now, and Congress/the admiralty have been paperig over leadership failures by spending more money, and blaming technology
• Animats 5y ago
  "Because of a UI issue, they couldn't figure out where throttle control was and ran into someone." That's happened more than once. NYC ferryboat allision.[1] It's another case of touchscreen mania. With ship controls, it's common to have more than one control station for basic helm and throttle. When docking, it's common to drive from a control station where you can see the dock. So which station has control is an issue. Some systems have physical feedback, so that the wheel and throttles at all stations move together. That avoids mode confusion. Aircraft people get this, but few others seem to. [1] https://www.workboat.com/passenger-vessels/seastreak-ferry-accident-due-to-captains-error-says-ntsb https://www.workboat.com/passenger-vessels/seastreak-ferry-a...
• scarier 5y ago
  The oldest aircraft navigational aids to avoid hitting the ground are fairly simple, and predate software (the radar altimeter is a good example, which in its most basic form just tells you your altitude above the ground and works really well in most flight regimes). Ground proximity warning systems are more along the lines of what you're thinking, and rely on the aircraft's flight computer to predict an intersecting flight path with its digital terrain model within a specified period of time (typically relies on GPS and/or INS data). As long as the software isn't integrated into aircraft flight controls (auto-GCAS in fly-by-wire military aircraft, for example) it doesn't necessarily need super-rigorous testing, although when it is connected to automated flight controls it should* have* (looking at you, Boeing...) an extremely rigorous testing process to ensure it behaves deterministically and as the pilot expects under all normal and emergency conditions. This generally involves a lot of bench testing, with a (potentially very large) number of test flights to ensure it works as designed in enough distinct conditions to have full confidence in the system. TCAS (the family of systems aircraft use to avoid hitting other aircraft) is much simpler and doesn't rely as heavily on GPS/INS data, but generally only gives data on cooperative aircraft transmitting transponder or ADS-B data. A very simple version gives a traffic alert when another aircraft approaches within a specified distance of you, but may not give any other useful information (relative altitude/position, heading, etc.), and the most basic version I can think of just gives a range to a single cooperative aircraft. I think you'll find that you're off by a few orders of magnitude on the number of submarine navigators out there, and there are few enough certified avionics systems that most are used by a substantial enough audience (and, more importantly, mandated by enough national aviation authorities) to be worth the R&D effort. That said, military hardware is often decades behind the civilian world when it comes to many of these mandated safety features, so I don't find it surprising that military ships (often with proprietary radar and sonar systems designed primarily for combat, and not wanting to broadcast their location to potential adversaries) don't have good integration with commercial collision avoidance software designed to work in a collaborative environment with off-the-shelf hardware that may have only been installed as an afterthought, if at all.