Homicidal autopilots

Too many variables, too many components in the equipment, so many manufactures for each component, circuit and chip, different suppliers for raw materials for each chip, supervisors not looking, quality control engineer on a break, so much greed and skimping on quality and reliability, improper maintenance, so little care what happens to others ..

The end result is there is just too many variables for you to take it for granted the chance of something going wrong is too high so ...
Keep your eyes in the back of your head, keep your eyes open ...
U never know when something will fail ...

Just you tube autopilot error or crash ...

Yes im paranoid but if seen too many accidents and problems not to justify my paranoia ...

 

I was just speaking to a couple of Captains from a well known large fleet owner and they told me each Captain has to get written permission to use the autopilot on his vessel..ouch!

 

Ouch ja,

I am not aware of a failure of a "Anschütz" AP ever. (Thats the standard in the commercial fleet)

And I personally don´t buy that story.

 

I guess autopilots are like boats you have the crap variants and the more respectable good quality ones ...

 

I tend to agree that it is unlikely that a Captain has to have "written permission" to use the installed Auto Pilot.

I can imagine though, that in today's work environment where everything has to be Safety Analysed every day, or shift, Marine Officers will be signing a sheet that stipulates the procedures they must follow so that they are always in control.

 

There has been too many accidents caused by having the autopilot on and not by it failing hence the company policy.
Just having the autopilot on does not guarantee good seamanship, generally the opposite.
Just google accident with autopilot most of which would not have happened if the autopilot was off!

The company I am talking about has over 600 offshore vessels

 

Was there a common brand in the failures? or where they all from different brands ?

 

Vulkyn
There are some very good brands out there like the Raytheon/Anshuntz as Apex pointed out. Like a radar an IMO type approved autopilot needs to be very good.
What I am saying is autopilot related accidents are just about all user error hence I understood when the Captains told me they now have to get written permission to use them from head office.

 

Well, thats a different animal, and I fully agree! Though the topic was failure of AP units (which the thread opener has absolutely NO clue about BTW.)

Regards
Richard

 

In the early 80's I worked as engineer for a firm that custom built several autopilots based on gyrocompass output and a rudder indicator they were used extensively with no problems in many coastal vessels.


The electronic microprocessors that came along around that time replaced the analog circuits could be more prone to faults in the circuitry and more problematic were bugs in the programs. But after the initial shakedown they were reliable too. When I'm on a bridge now and look at the computer screens and the joysticks and the pc based navigation and autopilot I really wonder how reliable the software is and at what point some bug rears it's head.

It would be so easy for the software writers to add some watchdog code that had a bit of intelligence it's done on aircraft.

 

They tested the Space shuttle OS for 8 years before they used it......
I agree PC's on the bridge are very scary considering that the IMO and all classification societys have no clue how to deal with them.
Where else on a vessel can you have a power source just plugged it????
Not to mention hard drives that crash when the typical AHTS is trying to move something that wont move. ( HDD's which are vented so they also corrode)

 

It's not as easy as you might think and I speak as a software developer with over 30 years of experience. I quite agree with you about bugs in the code, however, but this is endemic to all software. There are the bugs that you know about, and those that have yet to manifest, but there are always bugs.

When we designed & wrote the datalogging system for our icebreaker, we deliberately did NOT use it to control the autopilot, though we could have. Instead we had a separate chartplotter that we wrote, displaying data from our instruments and processed by our software, showing where we wanted the ship to go and transferred waypoints to the commercial chartplotter software and left the autopilot strictly alone.

PDW

 

A good point, PDW, and one that is easy to forget in this age of slick, works-out-of-the-box gadgetry.

QA for software is very, very difficult. A safety-critical system can't be cobbled together in whatever way happens to work; redundancy, fault tolerance, error checking and recovery, etc. have to flow from the original design philosophy that formed the basis of the first block diagram of the system.

For some reason, we seem to expect something "new" and "innovative" every few months. As was mentioned earlier, the space shuttle's computers took years to develop- likewise for the avionics of most modern airliners. The recreational electronics guys don't have that kind of time to wait between releases, so something has to give- and it's a hell of a lot easier to tack together repurposed code snippets and libraries you already have, than to write a safety-oriented, fail-safe system from the ground up.

 

Actually there _are_ techniques and tools out there which make very difficult (albeit not unimaginable) to make programming mistakes. But those techniques and tools are not mature, mostly because they are not used by developers.
If you are not in the industry, you would think that IT have progressed a lot lately. This is not the case with software. We may have very sophisticated hardware (and HW folks do use techniques and tools which could be used for software with minor modifications), but unfortunately most of the programmers are just dumb to have a big picture on the issue, or under management pressure to deliver something fast, no matter how buggy is it, or both.
This is much like what I am doing in boat design, but they do think they are professionals, act accordingly, and - this is the most sad part of the story - even the most incompetent software companies like Oracle and Microsoft earn orbital amounts of money every year.