In the robotics community, the stuff coming out of George Hotz has always been considered a kludgy mess, and unsuitable for serious work. Dude is a talented hacker, but the idea that this will replace ROS is kind of a joke.
362 karma · joined November 30, 2021
It becomes presumptive to assume that Lisp is the be all, end all of car networking woes. Even if you accept that a tool to manage and organize computational tasks using a more functional paradigm would be helpful, that doesn't mean that a somewhat archaic, syntactically maligned language is the answer.
ROS tackles this somewhat yeah? But it's still very much a research project, despite its use in some parts of industry. The most significant development for computation in vehicles in recent times is still the automotive Linux efforts, so we are not really even in the ballpark to discuss specific languages.
Maybe you could make a model that sizes motors correctly per application? But you are still much better hiring an engineer that knows what they are doing that can explain what is going on and trouble shoot things when they go wrong. At a certain point you are trying to figure out how to completely replace an engineer with a machine learning model, which I would like to think is a bad idea.
To me, that is what makes this hack so noteworthy. Hacker got access, and then kept playing the long game. Very spooky. And yes, someone capable of this probably has a plan for gaining network access if they had a target in mind. Even with best security practice, this rabbit hole goes pretty deep.
The situation that arises into a very complex n-dimensional problem that you would want AI to search through is the coordinated motion of multiple actuators to achieve a very complex output. Like, picking something up of unknown weight, running while carrying it up a steep hill, waving it around while doing all this. We take it for granted as humans with brains that can perform all this stuff trivially, but it is extremely complex motion.
My general experience with other commercial offerings for simulation... is not great. In my experience, people usually end up migrating to gazebo, but I have been away from the field for a while now so it could be different. It is probably a situation where Nvidia will have a few coporate clients that they prioritize, and you are on your own to get it set up if you aren't on that lists. Pretty normal.
But beyond that, the kinematics as well as the force dynamics for controlling a serial manipulator are very well understood. So there aren't too many gains to be made by AI. It is difficult to implement in software due to some tricky situations about the nature of motion planning. Discontinuities around orientation approaches in 6-DOF systems for instance. But widespread use of serial manipulators is proof that, although challenging, they are relatively solved. It is always interesting to watch an AI model or genetic algorithm do some path planning, but this is a pretty well trod area of research at this point.
Now, when you want a robot to walk and pick things up at the same time... that is when AI becomes something to consider in order to figure out how the dynamics should work.
The main benefit of port knocking is that it allows you to present your ports as normally closed. If you have a system and you are worried about it getting port scanned for whatever reason, it makes sense to have a scheme where the ports are closed unless it receives a knock. So if someone gets access they shouldn't for whatever reason and pulls a portscan off, the information they get about your system is somewhat limited.
In this scheme, it would be much better to have an access point you authenticate with and then that handles the port knocking for the other devices. So it is a kind of obscure method that is really only useful in a specific use case.
As far as SSH keys go, I would argue that SSH support is so ubiquitous, and SSH access is so powerful that it is a reasonable security tradeoff. I also don't think that SSH ports should be exposed to the internet, unless it is a specific use case where that is the whole point, like GitHub. I'm very down on connecting directly through SSH without network access. The xz situation has validated this opinion.
I personally don't know any application where you really need supreme admin access to every device, from any device, from anywhere in the world, while at the same time it has extreme security requirements. That's a pretty big task. At that point, constructing a dedicated, hardened access point that faces the internet and grants access to other devices is probably the way to go.
No reason to go crazy with your home server if it is only you using it. You can get away with a crappy laptop that your roommate is done with.
Generally though, It would be pretty trivial to add noise for this stuff. You would probably want some noise on the orientation if you wanted to model more of a camera shake. Perlin noise is my go to if I want to introduce randomness to positions really quickly.
A really good way to tell is to consider the relationship between your input and output. PID get's thrown around all the time because a lot of the time you are talking about controlling an accerlation (gas pedal, throttle) to effect a position, so you theoretically need a PID to stand in for a 2nd order differential equation since position and acceleration are two derivatives apart with respect to time. But as you pointed out, you usually can generalize your input and not worry about the D term.
In the video game I'm making, I don't really worry about derivative control, but PID control is mostly there to give the illusion that someone is driving a vehicle rather than it be controlled perfectly by a computer program. I don't want to model out how "skilled" the driver is - all I have to do to give that impression is just give the vehicle better handling.
Everyone is happy! Especially the people who actually have to use the software made this way! As we all know, enterprise software has a reputation for being extremely user friendly...
I am planning an overhaul in a bit. I looked into sun, but would probably rather use git lfs honestly. There are also options for uncompressed files to be stored as text, but it doesn't seem like you can get away from binary data completely. I haven't tried to hard to focus on it yet. It is somewhat frustrating that there isn't a clearly good free option.
I'm one of those weirdos who actually likes git LFS, but only because I admit all it's short comings. But it is working really well for my game dev project. Not what I use to manage my asset library though. It is what I use to sync the most up to date binary files, which is really the use case it is maintained for anyway (think syncing build artifacts or large data sets).
It is weird seeing my hobbyist game dev and professional enterprise software world collide like this. Articles like this made sense when Facebook was willing to put up billions for VR services. But that was a bust, and now you have tons of consultants spinning their wheels. There is no reason any game studio or publisher needs to be told what game engine is good, I don't even think they care as long as they can make a good game. It's not a huge organization that you are trying to rally around some kind of decision.
This is why enterprise IT is a mess. Every time you need to do anything, you have to triple check that you aren't violating some clause buried in some obscure legal agreement that you probably don't have access to. Or else, you could cost your firm a quarter billion dollars or more. So you end up with reams of dead software that is unusable by design since it is being held hostage by various different commercial interests. I understand that is just the world we live in when millions of dollars are involved, but we can do better.
On the other hand I get the concern: you would want some legally enforceable agreement with TCS that they won't steal confidential information you share with them in good faith to steal business. Nor go and hire a bunch of staff to poach your contract. But documentation of the software that Transamerica paid for is secret, are you kidding me? They would have been fine, legally, if they got the materials directly from Transamerica. Because they got it from someone who merely used to work there, it is a quarter of a billion dollar mistake. Seems like a pretty narrow difference to me, hardly some kind of grand ethical quandary.
In this case it is very straightforward legally if they copied documents over. You just can't do that, full stop. But I'm uncomfortable to say that this is an equitable relationship between all these companies, because it gives a level of protectionism to DXC to do a really poor job in delivery knowing that the American court system has their back if they lose on re-compete.
But even if you want to torture this analogy, LLMs are a godsend to techncial service Companies. It's like saying "Oh, planes are faster so I guess people won't be spending as much time traveling and there will be less of it now".
What did you say? That it is also complicated to deploy properly? And there is no objective way to measure its success? Christmas keeps coming early.