I understand that part. But not the one about the cruise car almost hitting you.
So: one lane street, person in front of you turning left, you pass them to the right in the space where there might be street parking which extends into the intersection. During that time, the car trying to turn left is finally able to do so and the car immediately behind it is the cruise car. Cruise-car is now able to go straight and almost hits you as you're both competing for that one lane?
Nope, cruise car was in the other lane turning left. It turned into my lane just after I passed the car turning left and the cruise car made a sudden stop in the wrong (my) lane. Had it continued, it would've collided with my rear wheel - so I'm like 75% sure I wouldn't have been at fault.
So, by the sounds of things, everything worked as designed. In many similar situations with a human behind the wheel, that would have been a collision.
Based on personal experience (as a firefighter/paramedic), I wouldn't be surprised if a not-insignificant percentage of collisions (both car/car and car/bike) in SF are caused by that very scenario.
Personally, that sounds like an anecdote very much in favor of automated cars.
You're completely off base here. I'd say I do this 10-15 times a day. So ~100 times a week, and there is never an issue. I am infrequently the first car to do it. I put it in quotes because it is a very common practice in SF. You would have to wait through every light twice to get anywhere in the Mission if you didn't do this. It's mostly single lane with no turn light. I noted that it was odd because I very rarely have problems doing this.
Not entirely related, but the pain of the "paper cycle" as a "customer" of health-care is real.
It seems as if to avoid dealing with hipaa regulation it's all paper, fax machine, and come pick up your x-rays burnt on a cd during your work hours (cd that you have to buy).
Having not grown in the states, this type of practices seems very primitive to me. Amongst many other aspects of the American healthcare, but that's another debate.
I am actually not opposed to the "CD you have to buy" policy. If I was hospital network admin I would not want people just plugging in random usb sticks into my network.
I think the expectation is that these organizations should be able to send this data to each other. That they was put the data on a disk and then give that to a patient who the drives it over to another part of town and hands the disk over is a ridiculous waste of everyone's time.
It also highlights how far the entrenched vendors and hospitals will go to keep their customers "locked in."
And in most cases they can send that data to each other, at least in the sense that their existing systems are capable of interoperating based on open standards. But to make it work the provider organizations often have to configure and test data interfaces with other organizations. That's an expensive process and no one wants to pay for it.
It's also fun to introduce e as a matrix operator for 3d rotations.
It's useful for kinematics and having compact representations for axis angle notations. It's a little far in my head but at some point it felt like a ha-ha moment with the Euler identity, in the "2d version".
(At least if you’re ever needing to compose rotations; to apply a quaternion to a big array of 3-vectors, go ahead and convert it to a 3x3 matrix first, which should end up slightly more efficient.)
3x3 rotation matrices are really hard to keep normalized properly, whereas quaternions are trivial to normalize (just divide by the norm).
If you need to compress a unit quaternion down to 3 numbers, instead of taking the logarithm (which is computationally expensive and a pain to deal with) use the stereographic projection.
Quaternions are a useful tool for manipulating rotations in a lot of common applications. But that wasn't my point here. Also quaternions are hard to grasp by humans. I find axis-angle much more palatable in general.
Imaginary numbers can be represented with a 2x2 skew symmetric matrix with no stretch of the imagination at all. And 3x3 skew symmetric matrixes represent rotations most compactly with only 3 actual variables. Instead of 4 for quaternions, 9 for "classic rotation matrixes", or the need to tell which is the order of the angles if you're given 3 euler angles.
There are interesting applications of Lie Algebra on SO(3) [1], notably in computer vision where a global energy is minimized across two successive rgb-d "shots" in order to recover the infinitesimal rotation [2].
It's going to be easier to minimize energy on something that is most compactly defined, and always amounts to a valid rotation.
From what I understand that would be (sorta; I’m not an expert in Lie theory) the logarithm of a rotation. I find the stereographic projection to be a more useful way to compress an arbitrary rotation down to 3 dimensions, for most purposes.
Yes, precisely. exp() is the mapping from so(3) to SO(3), so you can use log to go the other way around.
Here is the relationship with the other representations [1]
Where it becomes interesting for our rigid-body transform application (or recovery of it, in the case of computer vision), is with Twist coordinates (6 element vector) which will map to a 4x4 Transform, again using the matrix exp() operator [2].
I thought that too. Just finished a kaggle competition involving segmentation, like a lot of participant I used one form of U-net (my own implementation).
You can probably find a lot of u-net implementations from this contest.
One that performed really well [1]. It uses 'inception style' blocks feature extraction instead of vgg. But otherwise pretty similar.
I bought and assembled an ergodox some time ago. I found the key layout to be too esoteric for me to adapt my workflows to. The thumb cluster while a good thought is not as well positioned as on an Kinesis advantage to be extremely useful. I never ended finding a layout I liked.
It's been sitting in my parts drawer for a while, I'm not sure how to part ways with specialty items like that.
The kinesis freestyle + vip accessory was my daily driver for years but it's a bit mushy and the build quality is questionable (I went through 3 boards, including one DoA). My new one is the Matias Ergo pro. Quite similar, but better in almost every way, except price.
This is my daily driver and I really love it! It's mechanical, split and the layout is really not esoteric. The major upside from the UHK would be total firmware configurability, not an option on the Matias Ergo pro.
I would have to remap 'Mouse' though as this is the place for my 'Ctrl' key (I use Emacs a lot).
