Home
From time to time, I am called upon to explain some of the aspects of my research to friends and family. I'm working on drive-by-wire systems, I tell them, where there's no longer a physical connection between the controls of a vehicle and the devices being controlled. Almost always, I get a negative response back. Computers always crash, they tell me, how can you be sure your system won't just freeze and leave me without control of my vehicle at sixty miles an hour? (modern dbw systems are multiply redundant, certified by international safety organisations, provably safe, before they're ever considered for use on a vehicle.) In an emergency, the system won't let me pull the steering wheel as hard as I can and avoid the child/pet/tree that's unaccountably appeared in front of me. (First, good luck hanging on to the steering wheel when you're trying to command that kind of change in direction and speed, and second, most modern dbw systems won't let you get into that situation, never mind getting out of it.) It's pretty reliable. I've not yet had anyone see a drive-by-wire system in a positive light, only ever neutral or negative, despite the fact that we've trusted our passenger aircraft to identical systems for twenty-five-odd years now.
Well, I'm going to take advantage of a recent news article to present one of the positive sides of drive-by-wire technology. The technology is known as "platooning", or the "road train" concept, and it's being pioneered by a group of European engineers including members of a local engineering consultancy my lab works with now and then. Essentially, it involves a group of highly-skilled professionally-trained drivers running around the trunk roads of the country, driving modern drive-by-wire vehicles with a few custom modifications. Another modern drive-by-wire vehicle can then locate said driver and negotiate to couple on to the back of their platoon. The joining car disengages driver input, synchronises its control system with that of the platoon, and then pulls up to within a couple of inches of the rear bumper of the preceding car: from that point on, the entire platoon maneuvers like one long, flexible vehicle under the control of the lead driver. If and when a driver wishes to leave the platoon they can signal the system, which will increase the distance around their car, maneuver out of the platoon, then instruct the cars that were previously either end of them to close up together. At no point is there any physical connection: the entire system is electronic, wireless and RADAR based.
The potential advantages of this system are significant. First, because the cars are so close together, they behave aerodynamically like one vehicle, not several. This results in significantly less drag, and a fuel saving of up to 20% per vehicle (highest for those in neither the front nor back positions). Drivers in the platoon need not concentrate on driving while coupled, they can nap, eat, drink, work, talk to friends or business associates... Previous versions of the platooning concept required sensors and broadcast beacons to be integrated into the roadway, which is pretty unrealistic in the short and medium term, at least. SARTRE, the specific platooning system on which Ricardo are working, requires simply that the vehicles in the platoon have the required modifications, making it a system that could potentially be deployed nationwide in a few months with enough research effort and testing.
The Frankenstein complex will always be a problem for engineers in mine and allied fields, but it's worth bearing in mind that a lot of us drive cars using the technology we build. We're not always perfect, but we're not suicidal: if we trust it, knowing exactly how it works... maybe it's worth something other than a knee-jerk reaction?
Iain M Banks defines an Outside Context Problem as the kind of problem "most civilisations would encounter just once, and which they tended to encounter rather in the same way a sentence encountered a full stop." This, while a fascinating definition (and a rather good book) is not the kind of OCP I intend to talk about (briefly) tonight.
Mine's more mundane, but similar in the way it causes the brain to briefly stop processing while it figures out what's going on, and why it's wrong. It's meeting a celebrity at the supermarket checkout, it's taking a holiday to LongWayAwayLand and ending up going to the same bar as your neighbour. It's doing something in one place, setting and headspace, and having someone or something from quite another suddenly turn up. In this case, it was getting on the bus while decaffeinated and having a brief, somewhat strange conversation with a perfectly nice young woman you later (quite a lot later) recognise as someone who works in the Engineering cafe, and with whom you're perfectly comfortable chatting when you (a) are in the right place and (b) remember who she is.
I have no idea how universal these experiences are, so I figured I'd write about it and see if anyone has anything to say.
Ah, I have something to write about. That means I can put off the whole Big Post Of Doom until tomorrow, when I have time to write it.
People with whom I've been corresponding by email over the last few years will have noticed something rather peculiar: my emails are often squashed over to the left-hand side of the screen, with a line length of about 80 characters. Now, that's just fine: the email standard specifies a maximum legal line length of 76 characters (back in the day when you did everything at the terminal, that's all that would fit on screen, and no-one's seen the need to change it). However, in this era of wide-screen monitors it does look a little ... odd.
It turns out that GMail enforces the 76-char limit in its plain-text mode, by forcing breaks where they're not present. I do most of my email in plain-text mode, since I have little use for the extra features HTML email brings, but I hadn't noticed that there was any other editing option in GMail. On the top-left corner of the "Compose Message" box, there's a link that says either "Rich formatting" or "Plain text": whichever text you see, it's in the opposite mode. I've just flipped my account from plain-text to Rich Formatting mode, and the spurious line breaks have gone away (since it can do what it likes to the line-length of the HTML source, but the changes won't be displayed at the opposite end). This may or may not be a long-term fix, but it appears to have positive side effects.
If you're one of the N people who have responded "WTF?" to a slightly oddly-formatted email in the past, hopefully the problem has gone away. Enjoy :)
Just a quick one, while I wait for other things to happen...
Maybe you've noticed that Microsoft Excel behaves a little strangely when you alt-tab between its windows (not moving between documents in the way you expect, closing all the Excel windows when you close one spreadsheet, outright crashing sometimes, etc). If you have, well, you don't have to put up with it. It turns out that this is due to Excel trying to be a little too clever, and starting all the spreadsheets you open in a single copy of Excel rather than opening a copy for each spreadsheet, like most sensible programs. This small distinction does very strange things to its behaviour.
Fortunately, it's fixable. tech-archive.net tells you how. Thanks to Bluebottle, on IRC, for pointing this one out.
Have fun.
So, holiday is nearly over. I'm taking Monday as well in the hopes that I can fix my sleep pattern, as well as to compensate for the day I came in (Tuesday, when nothing useful was done for various reasons). It's still too soon: I want more downtime, but the work can't wait.
So it goes, and such seems to be life right now. I'm looking forward to this project being over, even if I've been working on it so long now I suspect I'll have trouble switching to something else. Diversity is good, and all that.
Still, content... I'm sure I meant to write something mildly interesting when I opened this window. I suppose I'm doing some tidying up... Ah, yes. Slackware 13.0 is out. I use Slackware as my OS of choice: for the most part it serves me well, but a failed graphics driver update caused some trouble for me a while back. Vertical tearing, instability, the usual. Since Slack 13 is out, I figured, why not? A reinstall solves the problem, and gets me a KDE 4.2.4 desktop, which is probably going to be interesting, as I've never used KDE4 before. So, download ISO, burn, install.
The installation process was pretty painless, since I had my wits about me and gave it the correct information. The installer picked up the RAID volume holding my home directory without any trouble, etc, etc. It was all going so well...
Yes. ATI. Red-headed step-child of the Linux graphics drivers market. The company whose drivers cause untold woes for Linux users everywhere. I'm sure they're quite lovely under Windows, but next time round I'm just buying a damned nVidia card. It seems the latest release of the driver doesn't run under the configuration used in Slackware 13.0, at all: it crashes the window manager reliably, and locks the system about one time in three at the moment. Hence, it is currently uninstalled, and I'm back to one monitor and two or three frames a second OpenGL performance until ATI get their finger out and realise they need to fix things. There will be a bug report later, when I am calm and awake and so on.
Driver support. It's always the damned driver support. I wish I knew why so many companies get it wrong. Are we not paying customers, or something?
EDIT: And now the gas-lift in my chair has given up the ghost. I'm going to bed.