They’re back! The San Francisco autonomous vehicle hijinks, that is, as Waymo’s fleet of driverless cars recently took up the fun new hobby of honking their horns in the wee hours of the morning. Meat-based neighbors of a Waymo parking lot in the South Market neighborhood took offense at the fleet of autonomous vehicles sounding off at 4:00 AM as they shuffled themselves around in the parking lot in a slow-motion ballet of undetermined purpose. The horn-honking is apparently by design, as the cars are programmed to tootle their horn trumpets melodiously if they detect another vehicle backing up into them. That’s understandable; we’ve tootled ourselves under these conditions, with vigor, even. But when the parking lot is full of cars that (presumably) can’t hear the honking and (also presumably) know where the other driverless vehicles are as well as their intent, what’s the point? Luckily, Waymo is on the case, as they issued a fix to keep the peace. Unfortunately, it sounds like the fix is just to geofence the lot and inhibit honking there, which seems like just a band-aid to us.
From the “Tech Doesn’t Make Everything Better” department, we’ve got news of a vulnerability in high-end racing bicycles that opens up a new vector for cheating. While our bike has been sitting sadly idle for the last twenty-odd years, apparently shifting technology has changed a lot, to the point where high-end derailleurs are no longer connected to handlebar-mounted shift lever by Bowden cables but now have servos that are linked to the shifters via Bluetooth. Anyone with more than a few minutes of experience with Bluetooth accessories and their default “123456” passwords can see where this is going. While there are no specific instances of cheating detailed in the story, one can imagine the hilarity to be had with a Flipper Zero while sitting on the side of a road at a course upgrade. To be sure, there are other ways to cheat, but we’re not sure we see the advantages of wireless shifting that offset the risks in this case.
Only 94 percent? A recent study claims to have quantified business spreadsheet errors, finding that 94% have critical problems. They came to this conclusion by mining literature from journal articles dating back to 1987, but rather than looking for papers with associated spreadsheets and analyzing them for errors, they looked for papers that discussed spreadsheet quality assurance. So this is sort of a meta-study, which makes us doubt the 94% finding. Still, we’d say it’s a safe bet that there are a lot of spreadsheets out there with critical errors, and that spreadsheet abuse is pretty rampant overall.
They say that if you’re not looking for your next job, you’re just waiting to get fired. That’s pretty much a tautology since there are only two — OK, three — ways out of any job, but it’s still good to always be looking for your next opportunity. So you might want to check out eejobboard, which allows you to do a parametric job search in the electrical engineering space. Pretty cool stuff.
And finally, we don’t have any information on this other than what you see in the video, but we’d love to learn more about these hardware FFTs. The video shows two implementations, one using a Zync 7020 FPGA, and one that uses over a thousand 74HC-series chips to do the same thing. If anyone out there knows the OP on this one, we’d love to get in touch.
I can’t imagine running shifters off of coin ($$$) cells, tiny motors grit and gunk to save having anything outside of the tube frame. No cables inside either. If the shifters ran off a 18650 with wiring in an easy to charge set OK.
As a hobby and commuter cyclist, the electronic shifting makes sense in some situations. Sure there are downsides (batteries, wireless protocols, etc.), but there are good things about it as well (quicker and smoother shifting being top among them).
Also being a hardware engineer, it always amazes me what gets through QA at larger companies (Shimano, Garmin, etc.). A lot of the ‘security’ seems to be a security by obscurity, or based on old techniques. That said, I don’t think a 128 bit RSA encryption on a shifter is reasonable or needed. These companies have to walk a fine line between “it just works for the average Joe” and “it’s secure enough that it can’t get hacked easily” and that is a hard balance to get right.
The Waymo “Band-aid” for their honking cars nuisance actually sounds like the right answer. It really IS reasonable for a self-driving car to “tootle” when it detects another car backing into it, because in most situations, the other car isn’t also a Waymo car. The fact that the honking was a nuisance was because there were an inordinate number of cars maneuvering at the same time, at an hour that most people are sleeping. So geofencing solves the actual problem. I wouldn’t want the cars to not honk just because it’s 4 am, because the harm caused by a collision far exceeds the harm caused by one tootle. And while it might seem to be prudent to inhibit the tootle if the offending car is one that you already have a wireless connection to, this is a solution that would require a decent amount of validation.
+1. But why are the cars roaming around at 4 AM?