![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
purplecatLast week I went to a workshop on Formal Methods for Aerospace in Eindhoven.
The first (invited talk) was from Klaus Havelund of NASA's Jet Propulsion laboratory. During the course of discussing the verification of log files, he showed us the start of the following video showing the planned landing of the Mars Science Laboratory.
There is, he noted, a lot that can go wrong.
I was also amused by the second invited speaker's (Henk Blom) discussion of applying probabilistic modeling methods to look at a propose new European Air Traffic Control system in which flight path conflicts would be resolved locally, by pilots with assistance from instruments, rather than ground stations. Basically, if the pilots saw a plane ahead they would take action to avoid that plane without considering where other planes in the airspace might be in order to take a path that would also avoid them. He had managed to show that even with only 8 planes over an area the size of the Netherlands the probability of in flight collisions (given certain assumptions about the probability of various types of pilot error and equipment malfunction) was unacceptably high. As presented this looked a bit as though the designers simply hadn't taken airports into account, but a quick skim of the relevant documentation suggests that they were always proposing that ground control would continue to orchestrate things around airports. Even so, I can imagine he didn't make himself popular when he reported this back to people who had been working on the system for years, building expensives simulators and training pilots to use them.
The first (invited talk) was from Klaus Havelund of NASA's Jet Propulsion laboratory. During the course of discussing the verification of log files, he showed us the start of the following video showing the planned landing of the Mars Science Laboratory.
There is, he noted, a lot that can go wrong.
I was also amused by the second invited speaker's (Henk Blom) discussion of applying probabilistic modeling methods to look at a propose new European Air Traffic Control system in which flight path conflicts would be resolved locally, by pilots with assistance from instruments, rather than ground stations. Basically, if the pilots saw a plane ahead they would take action to avoid that plane without considering where other planes in the airspace might be in order to take a path that would also avoid them. He had managed to show that even with only 8 planes over an area the size of the Netherlands the probability of in flight collisions (given certain assumptions about the probability of various types of pilot error and equipment malfunction) was unacceptably high. As presented this looked a bit as though the designers simply hadn't taken airports into account, but a quick skim of the relevant documentation suggests that they were always proposing that ground control would continue to orchestrate things around airports. Even so, I can imagine he didn't make himself popular when he reported this back to people who had been working on the system for years, building expensives simulators and training pilots to use them.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
(no subject)
Date: 2009-11-09 02:22 pm (UTC)Did you have the workshop at the Evoluon?
And was Wubbo Ockels there too (a Dutch physicist and astronaut)?
(no subject)
Date: 2009-11-09 03:29 pm (UTC)"Unacceptably high" isn't as scary as it sounds...